No One Knows How Many Children Die in Custody

Did the state of Minnesota recently put a 16-year-old incarcerated boy to death? According to data it reported to the federal Bureau of Justice Administration (BJA) under the Death in Custody Reporting Act (DCRA), this execution took place on October 16, 2024. But the Supreme Court outlawed the use of capital punishment for minors in 2005, and the state of Minnesota itself abolished the death penalty in 1911. So why did the state of Minnesota tell the federal government that they executed a child? 

In truth, this state killing never happened. The report was a data error, as we could clearly see from other deaths-in-custody records we obtained from the state of Minnesota. Tragically, the death itself did occur; the child in question died by suicide inside a youth detention facility a month before Minnesota supposedly carried out his execution. It its reporting of the death, the only details the state got right were the boy’s name, age, and the fact that he died in custody. 

We point out these mistakes not to make light of a child’s distressing death, but to highlight the basic policy failure it represents. An internal investigation by the Minnesota Department of Corrections Office of Inspector General found that the youth detention facility where the teen died had violated multiple state standards leading up to the suicide, skipping mandatory wellness checks and hiring staff who lacked required qualifications. And when it came time to report the death to the federal government, the state couldn’t even get its facts right.  

Over the past year, the UCLA Law Behind Bars Data Project has been auditing the in-custody death data that states report to the federal government pursuant to DCRA. So far, we have reviewed data in 38 states, though only 33 provided data on the ages of the individuals who died. Between October 1, 2019, and December 31, 2025, those states reported the deaths of roughly 200 people 18 years old or younger—though the incomplete, dubiously accurate nature of the data means that the true number of youth deaths in custody remains unknowable. 

Errors and Missing Context

Taken at face value, the data raises plenty of concerns. As with Minnesota, Maryland also reported having executed a child, and Illinois and Ohio together reported half a dozen cases in which police allegedly shot and killed newborn babies. In Texas, the Houston Police Department told the state attorney general that its officers had been disarmed not once but twice by infants just days old; other Texas law enforcement agencies have supposedly faced down knife-wielding, motorcycle-stealing tots, while the Texas state prison system has reportedly had multiple babies die from drug and alcohol abuse. 

All these examples, of course, reflect errors in data reporting, most commonly the result of careless data entry processes that mixed up the date of death and the date of birth. Mercifully, the Texas attorney general caught the mistakes just noted before reporting the deaths to the federal government (though they didn’t bother to go back to the agencies in question and get the correct ages of the deceased), but Maryland, Illinois, and Ohio all reported these obviously erroneous datapoints to BJA. 

Even when reported deaths don’t contain such flagrant mistakes, the reports often lack critical information or contain misleading details. For example, five deaths involving children age 14 or younger (two in Georgia and one each in Illinois, Indiana, Massachusetts, and Maryland) are listed as “accidental,” and the states provided no other information about the circumstances of the deaths in the data they released in response to our records requests. However, using news reports, we were able to determine that at least three of these deaths occurred from vehicle crashes during police chases. According to BJA regulations, deaths during police chases should only be reported as accidental if they occurred without police intervention, but in at least one case, officers allegedly used as “PIT” maneuver, in which a police car intentionally rams a fleeing vehicle; that death should have been reported as a use-of-force incident under BJA regulations.

Bad Data, Real Concerns

Even with the flaws in the data we obtained, the reported numbers point to disturbingly high numbers of children dying in state custody or during encounters with law enforcement. Notably, 124 deaths were reported as “arrest-related deaths,” a category that includes things like police shootings and fatal vehicle pursuits. However, the true number is likely much higher. Not only are we missing data for many states, but a 2022 report by the congressional Government Accountability Office estimated that states failed to report nearly half of all arrest-related deaths, fueled in part by state reporting systems that largely rely on voluntary cooperation by hundreds of individual law enforcement agencies in each state. The Police Data Archive, which uses media reports to track deaths caused by law enforcement, estimates that at least 227 youth died during encounters with law enforcement between October 2019 (when the Bureau of Justice Administration began collecting in-custody death data), and the end of 2025. 

Additionally, we identified 22 deaths that likely occurred within youth detention facilities. The standard DCRA questionnaire does not include a specific category for youth detention facilities—it lumps them together in a catch-all group for “other state or local correctional facilities, including juvenile facilities”—and states did not consistently use this category even when other details, such as the facility name, made it clear that a death occurred when the child was in youth detention. For instance, Minnesota reported the supposedly-executed 16-year-old's death as occurring in a state prison, even though at the time of his death, he was incarcerated in a youth facility.  

Nonetheless, these deaths revealed disturbing patterns. Georgia was responsible for 16 of the deaths we identified, an average of three children dying in custody each year in a state that incarcerates fewer than 2,000 minors at any given time. Georgia’s youth incarceration system has serious problems. Congressional investigations, local media reports, and state audits of Georgia’s juvenile detention centers have revealed rampant physical and sexual abuse, unsafe conditions, and overcrowding, all of which may contribute to higher rates of deaths in custody. 

Yet do Georgia’s reported death totals reflect deadlier youth facilities or do they instead simply indicate better reporting practices? Given the wide variation in data collection practices across the country, there is no way to tell. Unlike many states, Georgia’s state administering agency (the state agency responsible for reporting deaths to the federal government under DCRA) has a data-sharing agreement with the Georgia Department of Juvenile Justice, meaning they obtain death reports directly from the agency responsible for running youth facilities. In many other states, state administering agencies rely on voluntary cooperation from youth detention agencies or else they collect data indirectly from coroners or medical examiners, with the effect that deaths can easily fall through the cracks. 

Forgotten Children and a Potemkin Data Collection

Last month, researchers at Georgetown University’s Forgotten Children Initiative published a report documenting more than 1,000 graves—many unmarked—of children who died in youth detention facilities across 29 states. Most of these children died decades ago, before any systematic official effort existed to track their deaths; the true number of deaths is likely much higher. But this problem has in now way been solved, and deaths of children in custody are still being forgotten and remain invisible today. 

For most of U.S. history, no formal efforts existed to systematically track the deaths of children in law enforcement custody. When Congress first enacted DCRA in 2000, deaths in youth detention centers were included among the categories of in-custody deaths that states had to report. Nonetheless, for the next 19 years, the Bureau of Justice Statistics only collected data on deaths in prisons and jails, meaning that the deaths of children in custody were only counted if they occurred in an adult facility. 

Congress reauthorized DCRA in 2014 in the wake of several high-profile deaths in custody, including the police killing of 12-year-old Tamir Rice in Cleveland, Ohio, less than one month earlier. Its supporters hoped that the legislation would bring much-needed transparency at a time when the deaths of Rice and others had shaken public trust in law enforcement. As Senator Patrick Leahy declared in a floor speech at the time, “This important opportunity for needed transparency comes at a time when many Americans are losing faith in our justice system. We are having an important conversation about the loss of human life in communities across the country. Here we have an opportunity to instill some measure of accountability and hopefully begin to restore some measure of trust in these communities.” 

More than a decade later, it is unfortunately clear that DCRA has failed to accomplish most of its basic goals. Not only did the Department of Justice drag its feet for five years before beginning DCRA data collection, but when reporting finally started, DOJ declined to hold states accountable for reporting complete and accurate data. As a result, DOJ itself estimated that at least 5,000 deaths in custody went unreported during the first three years of DCRA data collection. “This underreporting is widespread, and not the result of a small number of lagging or uncooperative states,” the DOJ wrote in a report in September 2022. 

Since then, signs of improvement have been few and far between. The DOJ has refused to release individual-level data collected under DCRA, but an inadvertent data leak by the agency in 2023 allowed The Marshall Project to identify numerous missing deaths and inaccurate data. The result is a data collection effort that costs states and the federal government millions of dollars each year while failing to answer the basic questions that motivated its creation. Like a modern-day Potemkin village, federal in-custody death data offers little more than a superficial performance of transparency and accountability. 

To unearth and correct errors in data reporting is extremely labor-intensive. To find and confirm the mistakes we identified in this blog post required dozens of public records requests, hundreds of back-and-forth communications with state agencies, and countless hours of data cleaning, validation, and research. If state and federal agencies did their jobs correctly, those resources could have supported research into understanding why so many youth are dying in custody. Instead, we’re still struggling to figure out how many died in the first place.