Death at the border

Eschbach, Karl

Death at the Border1

Debates about United States border control policies have generally ignored the human costs of undocumented migration. We focus attention on these costs by estimating the number, causes and location of migrant deaths at the southwest border of the United States between 1993 and 1997. We document more than 1,600 possible migrant fatalities along the border in this period. More than 1,000 of these deaths were reported by United States data sources, and the remainder were Rio Grande drowning deaths reported by Mexican sources. Additional deaths may go unrecorded because the bodies of the decedents do not come to the attention of government officials. Deaths from hyperthermia, hypothermia and dehydration increased sharply from 1993 to 1997 as intensified border enforcement redirected undocumented migration flows from urban crossing points to more remote crossing areas where the migrants are exposed to a greater risk of death.

The end of the twentieth century witnesses a paradoxical development at the border between the United States and Mexico. Capital flows, commercial ties and labor streams that span the border have established a web of social and economic interdependency between the countries (Martinez, 1994). The ties between these nations grow closer as international boundaries continue to decline in significance throughout the world and as NAFTA leads to integration of the economies and societies of the United States and Mexico. At the same time, the United States-Mexico border is an intensified zone of conflict as the United States government escalates its efforts to seal the southwest border from undocumented migration (Dunn, 1996; Andreas, 1998). The cornerstone of the Illegal Immigrant Reform and Immigrant Responsibility Act of 1996 (IIRAIRA) – improvements to border and interior enforcement– funnels the lion’s share of enforcement resources to the United States-Mexico border, while only modest increases are being directed towards either the northern border or interior enforcement (U.S. General Accounting Office, 1996).

The situation in the United States is not unique. Unstable economic and political conditions in many regions of the world have ushered in unprecedented migration to the more developed areas (Castles and Miller, 1993). Efforts by developed countries to control global labor and refugee migration flows impose human costs. Seeking to evade tightened border control efforts, migrants frequently take substantial risks in their attempts to cross international borders, and some die in the process. For example, several hundred migrants are reported to have drowned in the past five years while attempting to cross the Strait of Gibraltar (White, 1997). From 1993 to 1997, one source documented 920 migrant and refugee deaths in the European Union; many of these deaths occurred to migrants during the crossing of an international border (United for Intercultural Action, 1997).

Undocumented migrants who attempt to cross the southwest border into the United States also confront a myriad of dangers (Durand and Massey, 1995; Chavez, 1992; Bailey et al. 1996). They cross the Rio Grande and other swift– moving waterways under the cover of darkness. They travel in sealed and poorly ventilated freight compartments of trains or trucks. They hike through the parched terrain of the American southwest. Some scale fences and other steel barriers erected by the United States government to seal the border. Endeavoring to overcome these multiple barriers, migrants often rely on the assistance of a coyote, exposing themselves to a criminal underworld. Tragically, these dangers sometimes lead to fatal consequences. In the period from 1993 to 1997, more than 1,600 deaths were recorded in the United States-Mexico border region in circumstances that indicate that they may have occurred during an attempt to enter the United States without inspection. Additional deaths may never have been registered because the bodies were never found by government officials. An uncertain number of would-be migrants die at the northern border of Mexico, some by drowning in the Rio Bravo (Rio Grande). Mexican records indicate that river drowning deaths exceed 500 between 1993 and 1997. Migrating without papers, many of these casualties remain unknown at death or are identified as “Mexican.” Collectively, they have become the desaparecidos (the disappeared) of the border.

Despite this tragic loss of life, mortality in the course of unauthorized migration has yet to be addressed by either international migration scholars or demographers. The bulk of research on unauthorized (Mexican) migration to the United States has been dominated by a number of reoccurring questions, including those concerning the decision to migrate (Binational Study on Migration, 1997), the size of the undocumented population (U.S. Immigration and Naturalization Service, 1997), labor market and public sector impacts (National Research Council, 1997), settlement and incorporation outcomes (Browning and Rodriguez, 1985; Hagan, 1998; Massey, et al., 1987), and the efficacy and results of different policy instruments for controlling undocumented immigration (Bean et al., 1994; Espenshade, 1994; Andreas, 1998; U.S. General Accounting Office, 1997). Indeed, current debate about border enforcement and undocumented migration along the United States-Mexico border has largely been limited to the effectiveness of border enforcement policies for curtailing unauthorized immigration (U.S. General Accounting Office, 1997).

In this article, we introduce the human rights issue of migrant mortality and unauthorized border crossing. Several questions directed our research and organize the findings presented here. How many undocumented people die while attempting to enter the United States? What are the causes of migrant deaths along the border? Where do migrants die along the border? Finally, what is the relationship between border enforcement efforts and migrant mortality? We begin by addressing issues of definition, measurement and undercount, and then lay out the research strategy that we developed to tackle these questions.


The enumeration of undocumented migrant deaths is fraught with problems of definition and measurement. In death as in life, information about undocumented migration is scanty and uncertain (see Center for Migration Studies, 1987, U.S. General Accounting Office, 1998), and must be inferred from data sources designed for other purposes (Heisel, 1979; Robinson, 1980; U.S. Immigration and Naturalization Service,1997). Migration status is not recorded on death certificates, nor is it systematically noted by officials responsible for the disposition of bodies of deceased migrants.

Three issues complicate the enumeration of migrant fatalities. First, status as an intended undocumented immigrant to the United States is not well defined within Mexico. An analytic accounting of the mortality associated with undocumented migration should include deaths from causes such as transport accidents, assaults or exposure on the northbound journey to the United States border, but these deaths will generally not be distinguishable as those of migrants. Second, deaths to migrants traveling to interior destinations within the United States may not be recognized as occurring in the course of undocumented migration because of the remoteness of the border. The practical difficulties of identifying all such deaths are probably insurmountable. Third, even in border areas where local officials are alert to migrant fatalities, it can be difficult to distinguish deaths occurring in the course of undocumented migration, deaths to United States-resident undocumented aliens, deaths of aliens legally present, and deaths of unidentified persons.


In order to investigate the magnitude of undocumented migrant deaths in 1993 through 1997, we limited our geographical coverage and focus to deaths occurring in counties near to the United States-Mexico border, from causes and in circumstances that indicate that the deceased may have been crossing the border into the United States without authorizing documents (see Fig. I.) Our research is intended to give information about the deaths of undocumented migrants that occurred during an entry without inspection into the United States. To this end, field researchers from the Center for Immigration Research (CIR) at the University of Houston have been touring the border, from Brownsville/Matamoros on the Gulf Coast to San Diego/Tijuana on the Pacific Coast, since 1995. Officials with access to administrative records concerning accidental deaths were asked to supply lists of these deaths for the five year study period, together with information that could be used to infer whether the deceased was an undocumented migrant in transit, including information about cause of death, location found (e.g., “banks of Rio Grande”), sex, age, date found, whether the deceased was identified or was a Juanita or Juan Doe, and nationality and place of residence.

In preparing tabulations reported in this article, we relied, whenever possible, on lists of probable migrant fatalities supplied by medical investigators in county medical examiners’ or coroners’ departments. These departments are best equipped to supply lists of undocumented migrant deaths for two reasons. First, they have legal responsibility to review all accidental and unexpected deaths that occur in their jurisdiction, and so they should have a comprehensive record of these deaths. Second, medical examiners seek to identify precise causes of death and investigators look for next of kin for corpses that are hard to identify. Medical examiner offices in each border county in Arizona and California either supplied lists of probable migrant deaths or allowed CIR researchers access to case files so that we could compile these lists ourselves. The Office of Medical Investigations in New Mexico likewise supplied a computer database, including case notes, for all possible migration– related deaths in the state for this period.

The state of Texas presented greater difficulties. El Paso County is alone among Texas border area counties to be served by a professional medical examiner. In less populous counties, cause of death is determined by a Justice of the Peace. Our field research showed that local agencies in Texas differ in the completeness of their records. Therefore, we ended up by relying on vital registration death records supplied by the Texas Department of Health. (We also use vital registration data to estimate deaths in a second tier of California counties that are adjacent to the border counties.)

Enumeration of deaths in Mexico that occur in the course of intended undocumented migration to the United States is an important task, but is beyond the scope of this study. However, because of the importance of the Rio Grande as a hazard to migrants where it forms the border between Texas and Mexico, we sought to enumerate drowning deaths whether they were handled by officials in the United States or Mexico. For this purpose, we turned to the coroners’ departments of local police agencies and fire departments that were charged with retrieving bodies of deceased migrants from the river. For the entire study period, the Centro de Estudios Fronterizos y Promocion de los Derechos Humanos (CEFPRODHAC) in Reynosa, Tamaulipas, has maintained a clipping file recording press reports of bodies of drowning victims recovered by Mexican authorities between Nuevo Laredo and Matamoros. We compared the resulting data series to more fragmentary data supplied by police and fire officials in the same areas and found a close correspondence to official records where these were available. Because the CEFPRODHAC series gives consistent coverage, we report these data where available.

The identification of migrant fatalities from medical examiner records is not always straightforward. Investigator case notes sometimes report special circumstances that make this determination easy (e.g., if a traveling companion or the family of the deceased confirms that death occurred during an attempted entry without inspection). Some deaths also may be readily identified as occurring in the course of undocumented migration because some migrants take risks that are rarely ventured by persons legally present in the United States (e.g., running across an interstate highway or swimming or rafting across the treacherous waters of the All American Canal in Imperial County California). On the other hand, medical examiner records do not contain complete information with which to establish whether or not a death has occurred in the course of undocumented migration, because medical examiners have no legal mandate to track migrant fatalities. We included ambiguous accidental deaths in our tabulations in order to estimate upper bound limits of possible migrant fatalities among cases in the coroners’ records.

Using vital registration data, we estimate migrant deaths by tabulating accidental deaths occurring to persons who are both foreign born and who reside in a foreign country We also include deaths of unidentified persons because it is likely that a large fraction of unidentified accident victims in border regions are undocumented migrants. Counts are made more precise using detailed cause of death information contained in the International Classification of Diseases (ICD9) codes to eliminate deaths that do not occur in the course of migration (e.g., drowning deaths occurring during recreational activities — ICD9 code E910.2 – are eliminated). Death totals reported from vital statistics data are upper bound limits of registered deaths of undocumented migrants that in the categories reported. Two exceptions are that in tabulating possible migrant deaths using vital registration data we excluded 1) deaths from natural causes and 2) deaths of foreign-born United States residents. Both of these categories will include a few deaths occurring during an undocumented entry, but the total number of such deaths would not be meaningful for the purposes of our study because a large majority of them will occur to persons legally present in the United States.

The likelihood that an accidental death that is reported in a vital registration database had occurred during an undocumented entry varies by cause of death. Table 1 reports the percentage of possible migrant fatalities listed in vital registration data for San Diego and Imperial Counties that were listed as probable migrant deaths by local medical investigators (individual cases were matched). The table shows that a large majority of deaths from train accidents, environmental causes (hypothermia and hyperthermia), and drowning were those of undocumented migrants. For other causes of death, undocumented migrants contributed a smaller but still substantial portion of the deaths. Of course, this proportion of deaths in any category that occur during undocumented migration could vary across different areas.


Our tabulations undercount undocumented migrant deaths in the border region. The most important reason is that officials do not learn about every death that occurs in their jurisdiction. One rancher and county commissioner in Texas observed of migrant fatalities in his own county that “however many deaths the sheriff tells you about, the true number is some multiple. There are millions of acres out there.” His sheriff agreed, saying “we don’t find the half of them.” A journeyman Border Patrol agent in Arizona confirmed this point for his own area, saying that “there are entries [by undocumented aliens] that we don’t know about; that means there are deaths that we don’t know about.” The discovery of these deaths can be especially difficult because a body lying in an open area in the Southwest may be dismembered by wildlife within a few days of death (Eschbach field notes, June 1998).

Rio Grande drowning deaths present similar problems of enumeration. Officials are frequently unable to find bodies reported to them. Brownsville fire department records show 36 such incidents between 1993 and 1995, and an incident log prepared by the Border Patrol for the El Paso area reported eleven unrecovered bodies between 1995 and 1997. Some of the bodies may have been found by Mexican officials. However, it is also possible some bodies are never found or float into the Gulf. In one incident in November 1996, eyewitnesses saw ten undocumented South Asian migrants swept off a sandbar by currents into the Gulf of Mexico as they attempted to swim across the river at its mouth (The New York Times, “Four Drownings along Border with Mexico Raise Alarms,” Nov. 24, 1996), but only three bodies were recovered by officials. Border Patrol agents in Brownsville told us that this was not an isolated incident.

There are other reasons that deaths are not discovered. Perhaps some are concealed by a smuggler who fears prosecution for homicide. The death total for persons locked in cargo compartments of trucks or trains cannot be known from official records. It is also difficult to count deaths from the delayed effects of an injury suffered during crossing, or an infection acquired by the migrant from a biohazard at the border, such as the polluted New River in Mexicali/Calexico, parts of the Rio Grande, or the storm sewers in Nogales, Arizona.

Some migrant deaths may not be officially registered even when they come to the attention of officials. Death registration is in general problematic in border areas (Shyrock and Siegel, 1979; Heisel, 1979). An element of discretion enters into the decision by local authorities to file a death certificate because of the state of decomposition of some of the bodies. Medical investigators in San Diego reported that they file a death certificate when they find skeletal remains of human extremities, and in Imperial County for “a body part that is essential to life” such as a skull or a spinal column. By contrast, one Texas sheriff reported that he would arrange for a death certificate on a human skull only if the decedent’s family became involved. Rural county officials in Texas, in particular, point to the cost of investigations, autopsies and burials for the migrants as a burden on limited county resources and as a disincentive to full formal investigation of the deaths of the undocumented.

What is especially problematic and tragic in this story of death at the border is that documentation of migrant mortality receives uneven attention by local and national bureaucracies. This results in an incomplete picture of the human toll of unauthorized crossing. It also makes efforts to develop migrant mortality rates associated with undocumented migration fairly meaningless. The denominator of such a calculation is problematic enough because flows must be estimated imprecisely (Espenshade, 1990); the numerator of the calculation is even more uncertain given a strongly suspected undercount. What is clear, however, is that the full mortality bill associated with undocumented migration to the United States is higher than we report if migrant deaths in all areas and from all causes were fully documented.


Table 2 reports possible undocumented death totals from United States sources by cause and location for the period from 1993 to 1997. Table 3 is a partial tabulation of Rio Grande drowning deaths reported by Mexican sources for the same period. Between them the tables identify just over 1,600 deaths for this period. The leading causes of death in United States border areas are drowning (29%), environmental causes (14%), motor vehicle accidents (13%), homicide (14%), auto-pedestrian accidents (9%), train trespasser or train pedestrian accidents (5%), and natural causes (4%). Other and undetermined causes account for the remaining 12 percent of deaths. Mexican data sources identify an additional 568 Rio Grande drowning deaths. California

San Diego County was the leading location for death at the border between 1993 and 1997, just as it was the leading port of entry for undocumented migrants in this period. By 1997, the shift of undocumented migration flows to other parts of the border seemed to be associated with declines in death totals in San Diego, along with increases elsewhere. Imperial County, California displaced San Diego County as the leader in migrant fatalities in 1997.

Deaths in San Diego are distributed among a large variety of causes, reflecting the variety of terrain types and hazards in that county. The leading cause of death is drowning, which accounts for 57 deaths in the period between 1993 and 1997. These deaths were concentrated in two periods in the spring of 1993 and 1995 when the Tijuana River was in flood. The Tijuana River is for the better part of most years a virtually dry stream bed that crosses from Mexico into the United States just before it runs into the Pacific Ocean near Imperial Beach. The river becomes a major hazard to migrants who attempt to cross it during its periodic floods.

Auto-pedestrian accidents were the second leading cause of migrant death in San Diego, contributing 50 deaths between 1993 and 1997. These deaths occurred primarily in two places. The majority of deaths occurred south of the Border Patrol’s checkpoint on Interstate 5 at San Onofre, north of the city of San Diego. Undocumented migrants exited northbound vehicles south of the checkpoint and were killed by southbound traffic as they crossed over to a strip between the highway and the ocean. Some of the deaths occurred near to the San Ysidro port of entry. Migrants crossed the border west of the port of entry and then tried to cross the interstates (I-5 and I-805) on foot to meet northbound rides. The number of auto-pedestrian deaths was reduced in the mid-1990s after the California state highway department fenced the Interstate median to prevent foot traffic and because the Border Patrol has been somewhat successful in deterring undocumented migration through western San Diego County since Operation Gatekeeper, an intensified enforcement initiative that began in this area in 1994.

When Operation Gatekeeper deflected migration flows eastward after 1994, environmental deaths become the leading cause of migrant death in San Diego county. These deaths are related to excesses of both cold and heat. Border Patrol officials report that as migrant traffic has increased in the remote and mountainous areas in the eastern part of the county, they are more likely to encounter migrant parties that are ill-prepared for the rigors of a walking journey that may last several days. Some migrants carry insufficient supplies of food or water, or are without the layers of clothing required by the notoriously “long thermometer” of the eastern parts of the county. Migrants who enter the mountains on winter days may experience unexpected sub-freezing temperatures at night. Deaths from natural causes reported for San Diego County are also typically environmentally-related deaths that occur in the eastern part of the county to persons who attempted to hike through this terrain despite preexisting health conditions that were aggravated by the exertions and conditions of the journey.

The sharp increase in death totals in Imperial County in 1997 is almost certainly a result of the deflection of undocumented flows eastward to this county because of the (correct) perception by the migrants that the Border Patrol has been less able to prevent undocumented entry here than in the better staffed and equipped San Diego sector. Imperial County stretches from the eastern boundary of San Diego County to California’s border with Arizona. The All-American canal claimed the lives of at least 40 migrants between 1994 and 1997. The other primary cause of migrant death in Imperial County is environmental exposure in its particularly harsh desert areas.


Relatively few deaths are reported for Arizona despite the desolation of its Sonoran deserts and the large increases in migrant traffic through the state after the implementation of Operation Gatekeeper. Some officials speculate that mortality totals are relatively low in part because migrants avoid the harshest terrain and because those who do attempt to cross are more likely to be prepared for its harsh conditions than in places where the terrain is less obviously forbidding. Another factor may be that the Tucson Border Patrol sector for the past several years has closed highway checkpoints during summer heat. This has the effect of reducing the length of the hikes that migrants take to avoid apprehension.

Migrant death totals did increase in Arizona between 1993 and 1997, though it is not yet clear whether this reflects an on-going pattern. In 1996, an unusually large number of exposure deaths were reported in the corridor west of Tucson, a pattern that did not repeat itself in 1997. In 1997, eight drowning deaths were reported in a single incident when a storm drain used by migrants in Douglas unexpectedly flooded.

New Mexico

Even fewer deaths are reported for New Mexico than for Arizona. Historically, New Mexico has not been a principal crossing area for migrants. However when border enforcement was intensified in El Paso in 1993, flows were redirected west to New Mexico. The principal crossing areas along the border with Mexico are populated, and migrants do not face the danger of dehydration and exposure in isolated areas that they do in other areas.

Texas and the Rio Grande

Nearly two thirds of the 1,950 mile southwestern boundary of the United States is defined by the Rio Grande/Rio Bravo between Texas and the Mexican states of Tamaulipas, Nuevo Leon, Coahuila and Chihuahua. Both legal and illegal crossing are concentrated near the densely populated areas of El Paso/Juarez (historically second to San Diego as a corridor for crossing), Laredo/Nuevo Laredo, the lower Rio Grande Valley, and to a lesser extent the smaller cities of Eagle Pass/Piedras Negras and Del Rio/Acuna. Intensification of enforcement near San Diego, El Paso and Nogales in recent years has funneled crossers to the lower Rio Grande through the Brownsville Corridor. This area has in turn become an area of increased border policing activities since the launching of Operation Rio Grande in the fall of 1997 (U.S. General Accounting Office, 1997). The mountainous Trans-Pecos region between El Paso and Del Rio is largely devoid of population settlement on either side of the river, and relatively few migrants are thought to cross in this area, though Border Patrol officials report expectations of increased crossing here because of increased enforcement elsewhere.

The greatest danger would-be migrants face along the Texas-Mexico border is the unpredictable flow of the Rio Grande. Texas vital registration data show approximately 40 drowning deaths of possible undocumented migrants each year. Mexican data sources show higher totals, averaging 130 to 140 for the three years where our coverage of the full length of the border is nearly complete. The river appears unimpressive in many spots because of the reduced flow due to irrigation diversions and a recent drought. Nonetheless, it does pose hazards to those who seek to ford or swim it. Dangers include sudden changes in depth such as from well holes under the stream bed, and vegetation along the river banks which can cause falls and catch the unlucky in its grip. In El Paso, irrigation canals that run adjacent to the main channel of the river pose dangers because of the swiftness of the currents and because in places the canals have been covered, preventing rescue of a migrant who is washed into the canal.

Rio Grande drowning deaths account for nearly half of the deaths reported in our tabulations. What portion of these deaths are those of local residents who die in the course of activities unrelated to border crossing? Thirty percent of decedents in CEFPRODHAC data are identified as residents of a Mexican community along the river. Informants in sheriffs’ departments in Cameron and Webb Counties in Texas, the medical examiner’s office in El Paso, and the police department in Juarez claim that the proportion of undocumented migrants among persons who drown in the river is high. Border Patrol officials express skepticism of this judgment and argue that further investigation is needed.

The incidence of Rio Grande drowning deaths fluctuates with river levels. River levels change both with rainfall totals in areas that replenish the river and with timed irrigation releases. In a period extending from October 1984 through September 1997, monthly counts of possible migrant drowning deaths recorded in Texas border counties were moderately correlated with river flow volumes as measured at International Boundary and Water Commission gauging stations along the river (R= .60 for El Paso County and .64 for Cameron County). Some of this correlation may be attributable to the shared rhythm of annual cultivation and migration cycles that places a large number of migrants in the river at the peak of springtime irrigation releases from the river’s reservoirs, but it is also likely that high water kills at a higher rate.

River flow volumes were at historic lows during our study period because of an extended drought in the areas that feed the river’s tributaries. For example, average river flow volumes through Brownsville, Texas, from 1993 through 1997 were only 30 percent of the average flow volumes for the preceding nine years. The average flow through Eagle Pass, Texas, for these years was only 62 percent of the flow for 1984 through 1992. The river was lowest in 1996 and 1997. As river flow volumes have been decreasing, migration flows have been increasing in Texas because of intensified enforcement elsewhere. The new Rio Grande migrants have been fortunate that they have encountered the river at an anemic ebb. The net result is that death totals reported by United States officials have remained relatively flat (Mexican data are harder to interpret because of concerns about the consistency of the data series and missing data points). The drought has probably reduced the number of drowning deaths in the river in the past several years below what would otherwise have been expected. If the river regains its former flow volume, we should expect an increase in migrant drowning deaths.

Additional deaths take place on the Mexican side of the river from assaults, falls and hyperthermia. We do not present tabulations of these deaths because we are only now assembling a comprehensive database about them. However, partial evidence suggests that these totals may contribute appreciably to the full total of crossing-related deaths. For example, municipal police and coroners’ departments in Nuevo Laredo report 22 deaths from falls or fractures along the banks of the Rio Grande in and near the city in 1997, and nineteen in the first part of 1998 (the banks of the river ascend in a steep incline to street level in Nuevo Laredo). A full accounting of crossing-related deaths requires attention to these deaths at the northern border of Mexico.

Hyperthermia deaths increased sharply in Texas between 1993 and 1997 just as in California. Much of the border region south and east of Del Rio is thinly settled mesquite-brush ranch land. The Border Patrol maintains checkpoints on the few highways leading away from the border to urban destinations in the interior. In several parts of the Texas border, migrants make extended hikes across ranches in order to circumvent the checkpoints. Dehydration/hyperthermia deaths associated with these hikes are reported in two areas in particular – along U.S. 77 north of Brownsville in Kenedy County and on trails that run north of Laredo and east of Eagle Pass.

Asked about the prevalence of migrant deaths in Kenedy county, the acting chief of the area’s Border Patrol station responded,

. . . in the past we would discover maybe one or two migrant bodies a year on ranch lands. We’ve never seen anything like this. . . . Several things are going on. First, more migrants are passing through here. In my opinion the concentration of federal manpower in El Paso and California is pushing them this way. Migrants who come through here usually have to rely on guides to bypass our interior checkpoints. Typically, groups of migrants skirt the check points by walking miles around them. The guide is anxious to get through fast, to the next drop off point where the groups of migrants will be handed over to another guide to take them north. If a migrant can’t keep up or gets lost, he is left behind to fend for himself. When the weather is hot and the conditions are drought, like they have been this past year, the dangers out there increase. (Hagan fieldnotes, April 1997)

As the Border Patrol agent’s comments summarize, several factors underlie fluctuations in migrant death. Weather, organized smuggling and federal enforcement activities all play a role in the number, location and cause of migrant border fatalities.


Table 4 shows the age, sex and nationality of the deceased. The majority of decedents (62%) are in their 20s or 30s. A large majority are male (85%). Age and sex composition of the decedent population reflect both the volume of flow in different age-sex categories and differences in risk-taking behavior by age and by sex.

The overrepresentation of males is consistent with the findings of the Zapata Canyon Project, which reported that in each year from 1988 through 1995, more than 90 percent of Mexican crossers in the study’s area were male (Bustamante et al., 1996). Women may be more likely than men to respond to the intensified enforcement by presenting false papers at official entry points, rather than attempting dangerous journeys over difficult terrain to enter without inspection – a speculation that is consistent with the findings of the Zapata Canyon Project that the proportion of female crossers declined in the early and mid-1990s.

The reported nationality of the decedents is overwhelming Mexican. It should be noted, however, that in the course of our fieldwork we found that in cases where bodies are not identified, Mexican identity is sometimes assumed for a body that appears to the untrained observer to be “Hispanic.” One Texas Justice of the Peace claimed to be able to discern Mexican identity by the clothes or physical features of the deceased (Bailey et al., 1996). The large number of cases without reported nationality reflect the large number of bodies found in border areas that are not identified, as well as limitations of official records used in the analysis.


United States border enforcement efforts were dramatically enhanced and restructured in the years between 1993 and 1997. The INS budget doubled from $1.5 billion in 1993 to $3.1 billion in 1997. Much of the increase has been directed towards policing the United States-Mexico border. From 1994 through 1997, INS spent more than $2.3 billion on border enforcement (U.S. General Accounting Office, 1997). A total of 31.7 miles of barriers were constructed on the southwest border between 1994 and 1997, and an additional 33 miles of fencing were either under construction or planned. The number of Border Patrol agents on the southwest border increased 76 percent between October 1993 and July 1997, from 3,400 to nearly 6,000 (U.S. General Accounting Office, 1997).

The Border Patrol has deployed its new resources at the border through a strategy that INS officials call “prevention through deterrence.” The new strategy is a two-pronged approach. First, intensified enforcement activities are implemented in stages along key points of unauthorized entry along the border. Second, to complement border enforcement, interior control is strengthened by increasing inspections along transportation arteries leading away from the border. The underlying logic of these border operations is to move unauthorized migration to rural terrain in which the Border Patrol believes that it has a tactical advantage. Specifically, migrants are channeled into areas more remote from major United States population centers, along more circuitous routes, and across more dangerous mountain and desert terrain, where the costs in terms of time, effort, financial resources and personal safety are higher. The combined effects of multiple risks, officials argue, is that the multiple barriers will eventually inhibit many from crossing, and if they do attempt to cross, the probability of apprehension will be increased (Andreas, 1998; U.S. General Accounting Office, 1997).

In 1993, “Operation Blockade” was deployed in the El Paso area; 450 agents working overtime covered a twenty mile stretch of the border. Apprehensions fell sharply and “Operation Blockade” (later diplomatically renamed “Operation Hold the Line”) was heralded as a success. In August of 1994, “Operation Gatekeeper” went into force south of San Diego, and in late 1994 “Operation Safeguard” was initiated in Nogales, Arizona. More recently, the Border Patrol has extended intensive enforcement eastward in San Diego County, and in August of 1997, “Operation Rio Grande” was launched along the Brownsville Corridor, extending from the Lower Rio Grande Valley to Laredo.

It is uncertain whether the United States campaign along the southwest border has decreased the volume of unauthorized entry (Bean et al., 1994; Espenshade, 1994; Espenshade, Baraka and Huber 1997; Singer and Massey, 1998; U.S. General Accounting Office, 1997). Enforcement activities have, however affected location, difficulty and mode of entry (U.S. General Accounting Office, 1994, 1997; Bailey et al., 1996; Andreas, 1998). Areas with concentrated enforcement have seen diminishing apprehension totals, reflecting the redirection of crossing attempts to other places. Areas with less intensive enforcement effort have had an increased volume of attempted entries and have sharp increases in apprehensions. For example, in the first six months of Fiscal Year 1993, the El Paso and San Diego Border Patrol sectors accounted for 68 percent of all apprehensions of undocumented migrants on the southwest border; by 1997 these areas of newly intensified enforcement accounted for just one third of all apprehensions (U.S. General Accounting Office, 1997:35).

Human rights activists fear that these border-policing activities have increased the mortality risks to the migrants. The hypothesis is that migrants respond to the selective closing off of familiar urban crossing points by adopting more difficult crossing strategies and by shifting to more dangerous desert, mountain and arid ranch terrain. Border Patrol officials and spokespeople generally acknowledge these new dangers, but suggest that responsibility for assuming the new risks lies with the migrants themselves and with smuggling operations that encourage migrants to take unnecessary risks. They also observe that urban crossing points have their own dangers and that effective border control in San Diego in particular has reduced migrant death totals there.


Is the border becoming a more unsafe place for undocumented migrants as enforcement intensifies after 1993? The evidence from the record of deaths from United States data sources is mixed. There is not a sizable overall increase in possible migrant fatalities in the five year period after 1993 as enforcement efforts intensify. Only about twenty additional possible migrant deaths are reported in 1997 compared to 1993. This is a small total number relative to the margin of error introduced by uncertainties about classification of deaths as migrant deaths and about the number of unreported deaths. Thus, it is impossible to conclude that there has been significant overall change in migrant mortality at the border.

When death totals are examined in detail by cause and location, a different conclusion emerges. It is unmistakable that death totals have risen dramatically from environmental causes in both Texas and California and from drowning in the All-American canal in Imperial County. Because of their circumstances, there is strong reason to believe that these additional deaths are in fact those of migrants and that the increases have occurred in part because of the deflection of migration flows from more populated crossing points to more remote and hazardous locations.

There are declines in some categories of deaths that partially offset these increases. The reduction that is unambiguously related to border enforcement effort is the decrease in auto-pedestrian accidents on San Diego County freeways after 1994. Officials with both the state department of transportation and the Border Patrol agree that the decisive intervention that has nearly – though not completely – eliminated these deaths was the fencing of freeway medians at the spots where most of the deaths were occurring. The reduction in drowning deaths in San Diego County after 1995 came about primarily because the Tijuana River (where most migrant drowning deaths happen in San Diego County) did not flood in 1996 and 1997. It is also possible that fewer deaths will occur in future floods because increased border enforcement in the Imperial Beach area has directed crossing attempts away from this hazard.

Reductions in homicide totals between 1993 and 1997 are hardest to interpret. It may be that decreases in homicides – particularly in San Diego County – are partially attributable to increased Border Patrol control of crossing points. INS Commissioner Doris Meissner has offered the formula that a border under control is a safer border (Joint U.S.-Mexico border safety press conference, June 1998). Parts of the border near San Diego are famous as places where border bandits once assaulted migrants with impunity (Wambaugh, 1984). These areas are now largely closed to both migrants and bandits. However, the hypothesis that enhanced border control is an important influence on the homicide victimization of migrants requires further investigation. Law enforcement agencies in border cities with enhanced Border Patrol enforcement efforts generally agree that increased enforcement has reduced crime where the enforcement is targeted. However, it is unlikely that Border Patrol activities account for all of the reduction in homicide in the region. Homicide rates dropped in many United States cities in this same period, and it is likely that the reductions in the border region reflect in part the effects of the causes responsible for national trends rather than increases in border enforcement.

Except for the decline in auto-pedestrian deaths on California freeways, the relation between decreases in fatalities and enforcement efforts is ambiguous, while the increases associated with the redirection of flows to south Texas and Imperial County are not. Thus it appears that the border was less safe for migrants in 1997 than it was before the onset of Operations Gatekeeper and Hold-the-Line in 1993 and 1994. Rising death totals from environmental causes are partly caused by drought conditions in the southwest in recent years. However, the deflection of migration flows has increased the proportion of migrants who are placed at risk of these weather-related hazards. On balance, rural crossing conditions have been more hazardous that the urban crossing conditions that are increasingly denied to the migrants.


The issue of migrant mortality introduces unfamiliar territory for migration research and raises important questions for policymakers. We begin the discussion by returning to our initial question: how many migrants are dying? Our research shows that the precise number of migrant deaths is elusive. In the absence of systematic recording of migrant deaths by a centralized agency, we found that local databases were partial and did not use common standards. Multiple agencies which kept records of mortality in a common area report different numbers of deaths. The discrepancies arise because of the different administrative purposes and definitions of different agencies in the absence of any responsibility to record incidents of migrant mortality.

The most important result of this lack of systematic record keeping about migrant fatalities is a shortage of information about the human costs of undocumented migration and a consequent inability of migrants, policymakers and the public to make informed responses to prevent the deaths from occurring. We have called attention to these deaths by offering estimates of one part of the death totals and by placing their occurrences into their local and environmental contexts. We emphasize that our own enumeration of 1,600 possible migrant deaths in a five-year period was necessarily partial because our coverage was restricted primarily to the northern side of the United States-Mexico border, and even in those areas it was hampered by the absence of the kind of organized recordkeeping that can occur only with official recognition that the deaths of migrants are a pressing public concern.

We welcome a fuller accounting that includes migrant deaths occurring throughout Mexico, as well as deaths throughout the United States, and at and beyond its northern border. Only the attention of the government of the United States and the international community can create the information base that will give a full understanding of the scope of mortality during undocumented migration and the changes in policies and practices that might save lives. We believe that INS, the principal state bureaucracy dealing with international migrants, should record these deaths. Such a systematic recording of deaths and injuries suffered by migrants is necessary in order to assess the full costs of border enforcement activities, as suggested by the U.S General Accounting Office (1997). Similarly, immigration scholars and demographers should include the human costs in their analysis of the effectiveness of border control activities.

By the end of 1997, it had become apparent that the deflection of flows to rural crossing points has elevated risks of environmental deaths for undocumented migrants and that increased migration through Imperial County in California has increased the number of deaths from a long-standing hazard, the All-American canal. While the final mortality bill had not yet been presented for 1998 by the time that this article was written, it is clear that the upward trend of these deaths accelerated in that year. If press reports are accurate, the death totals for 1998 substantially exceeded those for 1997. The summer of 1998 was long, hot and dry. The continued shift of migration flows to arid reaches of the southwestern landscape exposed migrants to the dangers attendant to the summer’s unusual heat.

Seeking to maintain a public profile of concern for migrant safety, the INS announced a joint safety campaign with the government of Mexico in the summer of 1998. The campaign includes Civil Air Patrol flights to search for migrants in distressed circumstances, advertising campaigns to warn migrants of dangers at the border, and targeting of enforcement efforts against alien smugglers who put their human cargo in danger. Border Patrol sectors have respond to concerns about migrant safety by adding four-wheel-drive EMT units, training field agents in emergency medical procedures, equipping agent jeeps with additional supplies of water and rescue equipment, and targeting additional patrols in dangerous areas during hot spells. Only time will tell whether the safety initiatives will reduce migrant death totals. It may be that some level of migrant mortality is inherent in the control of labor flows at the border. In particular, the heightened dangers of rural and desert migration may be an inevitable concomitant of border control that is increasingly effective at stopping flows through urban crossing points. Because human life is at stake, the issue requires careful and on-going study by both the INS and others who are monitoring migration and migration enforcement at the southwest border.

Ultimately, the most important questions may be ethical. How many migrant deaths are acceptable to the United States in its quest to enforce its borders? Do the local successes of strict border enforcement justify the mortality bill? Is the United States showing the same respect and concern for the value of the life of undocumented migrants that it does for its own citizens? The equality of all human lives requires that each of these questions be carefully considered when setting border and immigration enforcement policies. Moral obligations to respect human life should not be conditional on accidents of birth and citizenship, but are universal imperatives.

1Address correspondence to Center for Immigration Research, University of Houston, Houston, TX 77204-3474. This research was supported by grants from the American Friends Service Committee, the Ford Foundation, and the Trull Foundation. Karl Eschbach’s participation was supported by a faculty research fellowship from the Center for Mexican American Studies at the University of Houston. The authors are grateful to Hector Chavana, Aimee Krouskop, Jennifer O’Brien, Patrick Pierce and Anna Zakos for excellent research assistance, and to three anonymous IMR reviewers.


Andreas, P

1998 “The U.S. Control Offensive: Constructing an Image of Order at the Southwest Border.” In Crossings: Mexican Immigration in Interdisciplinary Perspectives.” Ed. M. Suarez-Orosco. Cambridge: Harvard University Press.

Bailey, S.

1996 “Death in the Border Zone: Causes and Processing of Undocumented Crosser

Fatalities. ” M.A. Thesis. Department of Sociology, University of Houston. Bailey, S., K. Eschbach, J. Hagan and N. Rodriguez 1996 “Migrant Deaths on the U.S./Mexico Border: 1985-1996.” Working Paper Series

96-1. Houston, TX: Center for Immigration Research. Bean, F. et al.

1994 Illegal Mexican Migration and the United States/Mexico Border: The Effects of Operation Hold the Line on El Paso/Juarez. Washington, DC: U.S. Commission on Immigration Reform.

Binational Study on Migration

1997 Binational Study on Migration: Migration between Mexico and the United States Mexico City and Washington, DC: Mexican Foreign Ministry and U.S. Commission on Immigration Reform. September.

Browning, H. and N. Rodriguez

1985 “The Migration of Mexican Indocumentados as a Settlement Process: Implications for Work.” In Hispanics in the U.S. Economy. Ed. J. Borjas and M. Tienda. New York: Academic Press. Pp. 277-98.

Bustamante, J. A., J. Santibanez and R. Corona

1996 “Mexico-United States Labor Migration Flows: Some Theoretical and Methodological Innovations and Research Findings.” Migration and Immigrants: Research and Policies. Mexico’s Report for the Continuous Reporting System on Migration (SOPEMI) of the Organization for Economic Cooperation and Development (OECD).

Castles, S. and M. Miller

1993 The Age of Migration: International Population Movements in the Modern World New

York: The Guildford Press. Chavez, L.

1992 Shadowed Lives: Undocumented Immigrants in American Society. Fort Worth, TX Harcourt, Brace and Jovanovich.

Center for Migration Studies and International Union for the Scientific Study of Population 1987 “Measuring International Migration: Theory and Practice,” International Migration

Review. Special Issue, 21(4):925-1613. Winter. Dunn, T. J.

1996 The Militarization of the US.-Mexico Border, 1978-1992. Austin, TX: Center for

Mexican American Studies, University of Texas-Austin. Durand, J. and D. Massey

1995 Miracles on the Border Retablos of Mexican Migrants to the United States. Tucson:

University of Arizona Press. Espenshade, T.

1994 “Does The Threat Of Border Apprehension Deter Undocumented U.S. Immigration? Population and Development Review, 20(4):871-92.

1990 “Undocumented Migration to the United States: Evidence from a Repeated Trials Model.” In Undocumented Migration to the United States: IRCA and the Experience of the 1980. Ed. E D. Bean, B. Edmonston and J. S. Passel. Washington, DC: The Urban Institute. Pp. 159-175.

Espenshade, T., J. Baraka, and G. Huber

1997 “Implications of the 1996 Welfare and Immigration Acts for U.S. Immigration,” Population and Development Review, 23:769-802.

Hagan, J.

1998 “Social Networks, Gender and Immigrant Incorporation: Resources and Constraints,”

American Sociological Review, 63:5547. Heisel, D.

1979 “Sources of Data for the Study of International Migration,” Quantitative Data and Immigration Research. Washington, DC: Research Institute on Immigration and Ethnic Studies, Smithsonian Institute.

Martinez, O.

1994 Border People: Life and Society in the U.S.-Mexico Borderlands. Tucson: Univerity of Arizona Press.

Massey D. et al.

1987 Return to Aan: The Social Process of Migration from Western Mexico. Berkeley CA:

University of California Press. National Research Council

1997 “The New Americans: Economic, Demographic, and Fiscal Effects of Immigration.” Washington, DC: National Academy Press. New York Times

1996 “Four Drownings along Border with Mexico Raise Alarms.” November 24, 1996. P. 17+.

Robinson, J. G.

1980 “Estimating the Approximate Size of the Illegal Alien Population in the United States by Comparative Trend Analysis of Age-Specific Death Rates,” Demography, 17(2):159-176.

Shyrock, H. S. and J. S. Siegel

1979 The Methods and Materials of Demography. U.S. Department of Commerce. Washington, DC: U.S. Government Printing Office. Singer, A. and D. Massey

1998 “The Social Process of Undocumented Border Crossing Among Mexican Migrants,” International Migration Review, 32:561-592.

United States General Accounting Office

1998 Immigration Statistics: Information Gaps, Quality Issues Limit Utility of Federal Data to Policymakers. GAO/GAD-98-164, July 31. Washington, DC: U.S. Government Printing Office.

1997 Illegal Immigration. Southwest Border Strategy Results Inconclusive: More Evaluation Needed. GAO/GGD-98-21, December 11. Washington, DC: U.S. Government Printing Office.

1996 `Border Patrol: Staffing and Enforcement Activities.” Report to Congressional Committees. GAO/GGD-96-65, March 11. Washington, DC: U.S. Government Printing Office.

1994 Border Control: Revised Strategy is Showing Some Positive Results. GAO/GGD-95-30, December 29. Washington, DC: U.S. Government Printing Office. U.S. Immigration and Naturalization Service

1997 Statistical Yearbook of the Immigration and Naturalization Service, 1996. Washington, DC: U.S. Government Printing Office.

United for Intercultural Action

1997 “920 Cases Documented: Refugees Find Death at the End of Their Journey. [webpage].

http://www.xs4all/~united/list.html. Wambaugh, J.

1984 Lines and Shadows. New York: Morrow. White, D.

1997 “Dire Straights of Eden,” Financial Times Weekend, August 30-31.

Karl Eschbach

Jacqueline Hagan

Nestor Rodriguez

University of Houston

Ruben Hernandez-Leon

Universidad de Monterrey

Stanley Bailey

University of California, Los Angeles

Copyright Center for Migration Studies Summer 1999

Provided by ProQuest Information and Learning Company. All rights Reserved