The number of children in foster care in the U.S. has fallen by more than 20% since its 2017 peak. Yet the disproportionate representation of Native American and Black children has persisted. For Native American children, in particular, disparities have grown steadily worse.
On a single day in 2024, roughly 345,000 children were in foster care in the United States — down from a peak of 440,000 in 2017, a decline of 22%. And while a reduction in the number of children in out-of-home care is, on its face, good news, not all of the foster care data tell quite as positive a story. The reality for Native American and Black children in the U.S. is that their disproportionate representation in the foster care system has held steady — and in some cases increased.
By comparing each state's foster care population to the racial composition of its overall child population, we can calculate a disproportionality index — a key indicator describing whether a group is over- or under-represented in the foster care system relative to the broader population of children. A value of 1.0 tells us that a group is represented equally in both foster care and the larger population. Values greater than 1.0 reflect magnitudes of over-representation. A value of 3.0, for example, demonstrates that children of that group are in foster care at three times their share of the child population. Similarly, indices with a value less than 1.0 reflect instances where groups are under-represented.
When we track that index over the past decade, the pattern is stark. As the foster care numbers fell, disproportionality for Black children barely moved — and for Native American children, it climbed almost every year, from 2.50 in 2013 to 3.17 in 2024. The system may have gotten smaller, but it did not become more equitable.
In the chart below, the bars show the total foster care population falling since 2017; the lines show the national disproportionality index for Native American and Black children over the same timeframe. If disparity simply tracked to the size of the system, the lines would fall with the bars. That was not the case — the Native American line rises while the overall numbers drop.
Total children in foster care (bars, left axis) and national disproportionality index by group (lines, right axis), 2013–2024. An index of 1.0 reflects parity with the child population. Source: Annie E. Casey Foundation KIDS COUNT (AFCARS), 2013–2024.
Before zooming into individual states, here is the national picture across five major groups. In 2024, two groups were over-represented in foster care and three were under-represented; the two over-represented groups, where the disparities are widest, are the focus of this brief.
| Race or ethnicity | National index, 2024 | Representation |
|---|---|---|
| Native American | 3.17 | Over-represented |
| Black | 1.76 | Over-represented |
| White | 0.86 | Under-represented |
| Hispanic or Latino | 0.80 | Under-represented |
| Asian or Pacific Islander | 0.12 | Under-represented |
National disproportionality index by race and ethnicity, 2024. An index of 1.0 reflects parity with the child population; above 1.0 is over-representation, below 1.0 under-representation. Source: Annie E. Casey Foundation KIDS COUNT (AFCARS), 2024.
The national averages hide substantial variation among states — and that variation has shifted over the last decade. Choose a group and drag the year slider, or press play, to watch the map change.
Disproportionality index by state, group, and year. Plum tones indicate over-representation, slate tones under-representation, and pale tones approximate parity. The scale caps at 5.0 — Minnesota's Native American index reaches 14.8, far beyond the top of the range. States left white have fewer than 10 children of that group in foster care that year — too few for a reliable estimate — so the white areas shift from year to year. Hover a shaded state for detail. Source: Annie E. Casey Foundation KIDS COUNT (AFCARS), 2013–2024.
Native American disproportionality is both the most extreme and the most geographically concentrated. Nationally, the index is 3.17, but the national figure can be difficult to interpret. The index excludes 22 states where the population is too small to measure. And in a handful of states, the gaps are vast.
| Highest state indices, Native American children | Index | Children in care |
|---|---|---|
| Minnesota | 14.76 | 1,140 |
| North Dakota | 6.51 | 527 |
| Wisconsin | 6.34 | 373 |
| Iowa | 5.81 | 74 |
| South Dakota | 5.04 | 938 |
| United States | 3.17 | 8,181 |
States with the highest Native American disproportionality indices, 2024, with the number of Native American children in foster care. Source: Annie E. Casey Foundation KIDS COUNT (AFCARS), 2024.
The pattern for Black children is the most consistent finding in the data. In 46 of 51 states, Black children make up a larger share of the foster care population than of the overall child population. The national index is 1.76. What varies is not whether the gap exists but how wide it runs — from more than fivefold in Wyoming to near parity in states with very small Black child populations.
| Highest state indices, Black children | Index |
|---|---|
| Wyoming | 5.43 |
| Utah | 4.44 |
| California | 3.69 |
| Wisconsin | 3.49 |
| New Jersey | 3.13 |
| United States | 1.76 |
The five highest state indices for Black children, with the national figure for reference. Source: Annie E. Casey Foundation KIDS COUNT (AFCARS), 2024.
Like with any social indicator, the disproportionality index has its limitations. The index should be treated as a starting point rather than a conclusion. Consider Minnesota and Oklahoma. Both have substantial Native American child populations and both have hundreds of Native American children in foster care — 1,140 in Minnesota and 502 in Oklahoma. Their disproportionality indices are 14.76 and 0.93, respectively. At face value, Oklahoma appears to have solved a problem Minnesota has not.
It is not so simple. A state's foster care data is not a complete census of its Native children, and Oklahoma is a case in point. It has unusually extensive tribal child welfare infrastructure — the Cherokee Nation, for one, operates its own Title IV-E foster care program and reports to the Adoption and Foster Care Analysis and Reporting System (AFCARS) directly rather than through the state, and Oklahoma courts share jurisdiction with tribes over most Indian Child Welfare Act (ICWA) cases. Some Native children are therefore served, and counted, outside the state system entirely. Differences in how a child's race is recorded can further obscure the numbers. The index can tell you that Oklahoma and Minnesota differ; it cannot, on its own, separate how much of that gap reflects a difference in practice from how much reflects a difference in who gets counted, and how. Oklahoma's near-parity number requires additional context before it can be trusted.
None of this means disproportionality lacks merit as a social indicator. Nationally, and in most states, the disparity is large, consistent, and holds across a variety of metrics (e.g., entries, exits, point-in-time counts). The index reflects the combined output of at least four distinct key factors — differential need rooted in poverty and housing instability; differential surveillance and reporting by mandated reporters; differential decision-making at the points of screening, investigation, and removal; and differences in how children are counted and their race recorded. The index cannot, on its own, tell you which one dominates in any given state. Two states can post the same index for entirely different reasons and require entirely different interventions.
Disproportionality data is often highlighted as though the number itself were the finding. This is not the case. An index that notably exceeds 1.0 establishes that a system is producing unequal outcomes for a particular group of children; it says nothing about whether the main contributing factor is poverty and financial hardship, reporting behavior in schools and hospitals, screening decisions at the hotline, removal decisions in the field, or otherwise. Those call for entirely different responses (e.g., concrete economic supports, mandated-reporter training, clear protocols and decision-support tools at screening, structured decision-making at removal, etc.), and investing in the wrong response is partly why inequities persist. Oklahoma is the cautionary case in the other direction: a state can post a number that looks like parity while the true picture is obscured by who gets counted and how.
So the important work begins using the index as a starting point and asking concrete questions. At which decision point does the gap first appear — in who is reported, screened in, investigated, or removed? Is race recorded consistently enough for the comparison to mean anything? How much of the gap would concrete economic support resolve, and how much reflects differences in how case-level decisions are made? And what happens after a child enters care — in reunification, time in care, and permanency? None of these questions are answered by the index itself, but the answers lie in administrative data, and can be analyzed at the decision points rather than only at the outcome. That is the difference between simply measuring a disparity and doing something about it.
One methodological point to note when reading the map: In states where very few children of a particular group are in foster care, the index becomes unstable — a handful of cases can swing it wildly. Following standard practice for this type of data, any group with fewer than 10 children in foster care in a state is displayed in white (unshaded) rather than given a value.
That threshold affects groups unevenly. For Black children, no states are excluded. The map excludes 22 states for Native American children and 17 for Asian and Pacific Islander children. The Native American map is genuinely informative across the West, Plains, and Upper Midwest, and unshaded across much of the South and Northeast — not because disproportionality is absent there, but because the populations are too small to measure reliably in a single year. Note that interpreting an unshaded state as "no problem" would be a mistake.
A note on the counts. The data in this brief come from the Annie E. Casey Foundation's KIDS COUNT Data Center — a Child Trends analysis of the AFCARS “AB” research files — and count children in care on the last day of the fiscal year. They run somewhat higher than the totals on the Administration for Children and Families' public AFCARS dashboard, which draws on a slightly different set of data files.