Methodology

As of August 2026, the M3P2 dataset includes reported migrant deaths that occurred from 1980 through 2023 on the United States side of the United States–Mexico border in California, Arizona, New Mexico, and Texas. Additional updates are planned for 2026–2027.

M3P2 documents the deaths of migrants—people who die while traveling to a country other than their country of residence. Our definitions of “migrant” and “migrant death” are based on those used by the International Committee of the Red Cross and the United Nations’ International Organization for Migration. Specifically, a “migrant” is a person (or persons) “who leave or flee their habitual residence to go to new places—usually abroad—to seek opportunities or safer and better prospects.” A “migrant death” refers to a person (or persons) “who have died at the external borders of states, or in the process of migration towards an international destination, regardless of their legal status.” [1,2]

We adopt these definitions of “migrant” and “migrant death” because they recognize both the motivations for migration (e.g., safety and employment) and migration as a process (e.g., journey, border crossing). We also expand our understanding of “migrant” from the International Committee of the Red Cross and the International Organization for Migration in two ways:

1. We explicitly recognize the geography of the United States—Mexico borderlands as a contributing factor to migrant deaths, including the physical environment (e.g., terrain and climate) and United States border enforcement policies, such as Prevention Through Deterrence [3]; and
2. We adopt a perspective that reflects the broader context of the United States—Mexico borderlands, recognizing migrant deaths as a regional phenomenon rather than as isolated incidents.

The M3P2 dataset is compiled from death reports obtained through public information requests submitted to state and county agencies along the United States–Mexico border. Primary data sources include medical examiners’ and coroners’ offices, county sheriff’s offices, and, in Texas, Justices of the Peace. In addition, the dataset incorporates one national-level source: publicly accessible unidentified persons reports from the National Missing and Unidentified Persons System (NamUs). NamUs cases were included only when the available case notes provided sufficient information to identify the decedent as a migrant. For example, a case would be included if the Case Circumstances notes stated, “Remains were recovered from a known migration corridor.”

The M3P2 dataset also incorporates data from No More Deaths, a humanitarian organization based in southern Arizona; Operation Identification, a program of the Forensic Anthropology Center at Texas State University; and the International Organization for Migration’s Missing Migrants Project. We also requested data from several other local, state, and federal agencies, including the United States Border Patrol’s Missing Migrant Program, but were unable to obtain access to their records.

A journal article coauthored by members of the M3P2 team and published in 2024 [4] explains the criteria used to determine whether a deceased person should be included in the M3P2 migrant mortality dataset. These four criteria are:

1. Context. The individual died in the process of migration, and the death is associated with the United States—Mexico border and United States border enforcement policies, including Prevention Through Deterrence. This criterion reflects a regional perspective that encompasses migration routes recognized by federal and local law enforcement agencies, humanitarian organizations, researchers, and migrant family and community networks.
2. Identification documentation. The absence of United States identification documents may indicate that the deceased individual traveled from outside the United States. When no identification is available, the decedent’s location (Criterion 3) and personal effects (Criterion 4) may provide evidence that the individual was traveling outside state-authorized channels or avoiding official ports of entry. Other forms of identification, such as a voter registration card, birth certificate, or non-U.S. identification card, may also indicate that the decedent was a migrant.
3. Geographic parameter(s). We identify geographic zones where migrants are more likely to experience border enforcement-related mortality during active transit. These zones include:
   1. Zone A: The area adjacent to the United States–Mexico border and in the vicinity of United States Border Patrol (USBP) permanent interior checkpoints.
   2. Zone B: The area between USBP permanent interior checkpoints and the USBP-defined 100-mile border zone.
   3. Zone C: Locations within the interior of the United States where there is clear evidence that the individual died in the process of migration [5].
      1. “Active transit” is an essential component of a Zone C classification. For example, the 50 migrants found dead in an abandoned tractor-trailer in San Antonio, Texas, meet this criterion [6]. In contrast, deaths associated with workplace enforcement operations conducted by the USBP or US Immigration and Customs Enforcement (ICE) would not qualify. Active transit cases and events, including the San Antonio case, occurring in the interior of the United States are not currently included in the M3P2 dataset because consistently identifying them across the full study period is not currently feasible, particularly given that the dataset extends back to 1980.
4. Personal effects. Items found on or with the human remains—such as foreign currency, religious objects, or other personal belongings—that indicate active transit and suggest a country of origin other than the United States.

This generalized guide to deaths during migration along the United States border facilitates consistent evaluation of past, present, and future records. These four criteria are not universally applicable to all migrant deaths. Our criteria are relevant to the United States southwestern border and might be informative for other fatal border crossing contexts. These criteria will help M3P2 and others ensure compatibility of data across United States-based jurisdictions as migrant mortality accounting and research continue.

All cases were reviewed and standardized before inclusion in the M3P2 dataset. When information was not provided by a source, the value “No information” was recorded in the corresponding field. The M3P2 dataset includes 26 standardized fields that capture key geographic and forensic characteristics of the decedent and the circumstances of death. This standardized structure facilitates comparisons of migrant mortality across the southwestern United States border region.

Certain data fields required additional processing, including “Date of Death,” “Land Access,” and “Landscape Type.”

The Date of Death field was used to create separate Year, Month, and Season columns. In the dataset, the original field name, Date Associated with Death, was simplified to Date of Death and should therefore be interpreted with caution. Dates were carefully reviewed by the M3P2 team because, depending on the case, they could represent the date the body was found, pronounced dead, or received by the medical examiner rather than the actual time of death. Consequently, the recorded date is not always the definitive date, or moment, of death. The recorded date corresponds to the actual time of death only when the death was witnessed, such as in the case of an accident or when medical intervention was provided.

Land Access and Landscape Type were assigned based on available location information. Land Access was deductively determined by the M3P2 team using location descriptions together with circumstance and scene-removal notes. Rather than describing legal ownership, this field characterizes the accessibility of the location or the types of individuals or agencies likely to discover the decedent. Four standardized categories were used: Public, Private, Military, and Tribal. Public locations include publicly accessible areas such as hospitals and military installations, whereas Private locations include privately owned land, lots, buildings, and construction sites.

Landscape Type describes the general environment in which remains were found (or, in cases such as hospital deaths, where the death occurred). Like Land Access, this field was assigned deductively using available location information. Five standardized categories were defined: a) Desert, including arid environments with little or no vegetation; b) Vegetated Area, including brushland, forests, parks, nature preserves, and agricultural land; c) Water, including rivers, canals, arroyos, dams, the Pacific Ocean, estuaries, washes, beaches, and embankments; d) Built Area, including lots, roads (trail, dirt, caliche, county, farm-to-market, street, avenue, boulevard, freeway, highway), construction sites, cemeteries, hospitals, and private residences; and e) Mountain, representing mixed rugged or high-elevation terrain.

Preparation of the M3P2 dataset included identifying and eliminating duplicate records, a time-intensive process because information for the same decedent was often obtained from multiple sources. For example, the same decedent might appear in more than one database or report, such as both a sheriff’s death report and a NamUs unidentified persons report. Record consolidation was therefore necessary whenever migrant mortality data were acquired from multiple sources within the same state or county. When multiple records existed for a single decedent, they were consolidated into a single record by prioritizing information from the agency responsible for the case, typically the medical examiner or coroner. Information from other sources was then used to supplement the primary record or resolve discrepancies, giving preference to the most recently updated information when appropriate.

M3P2 emphasizes the importance of positional accuracy because spatial analyses are only as reliable as the locations on which they are based. Positional accuracy refers to the degree to which a recorded geographic location corresponds to the decedent’s actual location of death or recovery. The precision of location information varies considerably among records in the M3P2 dataset. Some records include GPS coordinates collected at or near the scene, whereas others identify only a general location, such as a city, ZIP code, or county.

Consequently, positional accuracy reflects not only the precision of the recorded coordinates but also the confidence that they represent the correct location. Without qualitative and quantitative information on positional accuracy, the reliability of spatial analyses, including the identification of clusters, hotspots, and temporal patterns, is questionable. Accordingly, every available location record underwent a detailed manual review to verify its positional accuracy before being included in the spatial analyses.

Not all records in the M3P2 dataset include GPS coordinates (5,754 of 6,635 records). Each available GPS coordinate was manually reviewed to verify that it fell within the correct state and county and was consistent with other location information, including ZIP code, nearest city or town, and narrative location descriptions. Based on this evaluation, each coordinate was assigned a positional accuracy level. The table below summarizes the criteria used to assess positional accuracy.

The M3P2 team considered four factors when assigning levels of positional accuracy. First, the precision of the reported latitude and longitude coordinates was evaluated according to the number of decimal places provided. Second, the specificity of the narrative location description was assessed using location-related keywords (e.g., canal, survey marker, desert, and mountain). Narrative information was obtained primarily from scene removal notes and circumstance reports. When the geographic coordinates corresponded closely with the narrative description—for example, when mapped coordinates were located adjacent to the canal described in the scene removal notes—the location was assigned a higher level of positional accuracy.

Third, the final positional accuracy classification was based on the more conservative of the two sources of information: the geographic coordinates or the narrative description. High coordinate precision alone was not sufficient to justify a high positional accuracy rating if the accompanying narrative was vague or inconsistent with the mapped location.

Fourth and finally, the specificity of the narrative always constrained the final positional accuracy assessment. For example, a statement such as “remains found near a hiking trail” provides substantially less locational certainty than “remains found 1.5 miles southeast of the Vista Hiking Trail trailhead.” In other words, highly precise geographic coordinates cannot compensate for vague or absent spatial descriptions. The only exception involved Arizona cases, where volunteers from Humane Borders routinely communicate directly with the medical examiner’s office or sheriff’s office to verify reported coordinate locations, thereby providing additional confidence in their positional accuracy.

M3P2 supports humanitarian forensic action, defined as the application of forensic medicine and science to activities that alleviate human suffering and protect the dignity of victims, particularly in the aftermath of conflicts, disasters, and humanitarian crises. The project adopts the recommendations of the International Committee of the Red Cross, which are grounded in the four Geneva Conventions of 1949 and their Additional Protocols of 1977 regarding the treatment of the deceased and unidentified [7, 8]. In particular, M3P2 seeks to support “improving the communication, coordination and cooperation of forensic and investigative agencies involved in the recovery, analysis, documentation and management of decedent migrants […]” [9].

Beyond its research value, the M3P2 platform was developed to increase public awareness of migrant mortality along the United States–Mexico border and to provide an openly accessible resource for research, education, and humanitarian action. By making standardized, spatially referenced data publicly available, the project encourages users to investigate their own research questions, develop evidence-based policies and practices, and share their findings with broader communities.

A more complete understanding of the magnitude and geographic distribution of migrant deaths is essential for developing appropriate resources, improving training and operational procedures, strengthening forensic investigations, and ultimately reducing preventable loss of life.

[1] International Committee of the Red Cross/ International Federation of the Red Cross. 2009. Policy on Migration, available at: https://www.unhcr.org/ie/sites/en-ie/files/legacy-pdf/4bf646c89.pdf (Accessed 30 June 2026).

[2] IOM. Missing Migrants Project methodology, 2014, available at: https://missingmigrants.iom.int/methodology. (Accessed 30 June 2026).

[3] Homeland Security Digital Library, CRS Report to Congress, United States Border Patrol Operations: Prevention through Deterrence, 1997, available at: https://www.hsdl.org/c/view?docid=474656. (Accessed 30 June 2026).

[4] Miranker, M., Daniell, R., Kaplan, M., Flores-Guillen, V., Hernandez, J., Edgar, H., Spradley, K., Herrmann, N. and Giordano, A. 2024. What is a migrant death? An operational definition for a more accurate enumeration of migrant mortality along the US—Mexico border. Forensic Science International, 363(2024)-112156. https://doi.org/10.1016/j.forsciint.2024.112156. 

[5] Cuttitta, P. 2020. Preface: The increasing focus on border deaths. In P. Cuttitta and T. Last (Eds.), Border deaths: Causes, dynamics and consequences of migration-related mortality, 9-20. Amsterdam: Amsterdam University Press. https://doi.org/10.1515/9789048550203-002. 

[6] Edison, J. and Svitek, P. 2022. At least 50 people found dead in abandoned 18-wheeler in San Antonio. The Texas Tribune, June 27, 2022. https://www.texastribune.org/2022/06/27/bodies-18-wheeler-san-antonio-lackland/. 

[7] International Committee of the Red Cross (ICRC), Geneva Convention Relative to the Protection of Civilian Persons in Time of War (Fourth Geneva Convention), 12 August 1949, 75 UNTS 287, available at: https://www.refworld.org/docid/3ae6b36d2.html. (Accessed 30 June 2026).

[8] International Committee of the Red Cross (ICRC), Protocol Additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of International Armed Conflicts (Protocol I), 8 June 1977, 1125 UNTS 3, available at: https://www.refworld.org/docid/3ae6b36b4.html. (Accessed 30 June 2026).

[9] Cattaneo, C., Tidball Binz, M., Penados, L., Prieto, J., Finegan, O., and Grandi, M. 2015. The forgotten tragedy of unidentified dead in the Mediterranean. Forensic Science International 250: e1–e2.