Between the start of 2023 and the end of 2025 the US infant mortality rate fell 5.8%. That is real and it is good news. But split the first year of life in two and almost all of the improvement is in the later part — deaths between one month and one year fell 10.8%, while deaths in the first 28 days fell 3.0%.
Look at the annual figures instead of the rolling ones and the neonatal rate has been rising since 2021. This is the part of infant mortality that reflects pregnancy, labour, delivery and the first days of newborn care.
Each point is the twelve months ending that quarter, so consecutive points share eleven months of data and the line is smoothed by construction. These are provisional counts and will be revised slightly upward as late death certificates arrive.
Sudden infant death and unintentional injury — overwhelmingly postneonatal, and the targets of thirty years of safe-sleep and injury-prevention work — fell sharply. Congenital malformations, short gestation and low birthweight, and maternal complications — the causes concentrated in the first days of life — moved a few percent. Deaths per 100,000 live births; each panel on its own scale.
This is the whole national series and it has one shape. The newborn death rate fell in every single year from 2017 to 2021 — four consecutive falls, no exceptions. Since 2021 it has not fallen once: up, level, up. A four-year monotone decline reversing into a three-year climb is a stronger signal than any comparison of two years, and it does not depend on how the intervals are rounded. The same shape appears for Black infants (7.2 down to 6.4, then back up to 6.8).
Grey bar is the baseline year, rose bar 2024; the tick is the point estimate and the bar is the published 95% interval. Read the widths before the differences — American Indian and Alaska Native and Native Hawaiian and Pacific Islander rates carry intervals several times wider than the others, because the birth counts behind them are small. We have not marked any row “significant”: NCHS publishes these to one decimal place, which is too coarse to test overlap honestly. The claim on this page rests on the shape of the whole series, below, not on any single pair of years.
Neonatal deaths — the first 28 days — are dominated by prematurity, birth defects, and complications of pregnancy and delivery. Postneonatal deaths are dominated by sudden infant death, injury and infection. They are close to two different problems wearing one name, and the United States is currently making progress on one of them.
We are not saying newborn care has got worse. A rise of 3.5 to 3.7 per 1,000 is small in absolute terms, and some of it may reflect changes in who is giving birth, in gestational age at delivery, or in how deaths are certified. What we are saying is that the headline number — “infant mortality is improving” — conceals a component that stopped improving in 2021 and has moved the other way since.
It also sits beside something we measured separately: over the same period, admission to newborn intensive care among healthy full-term babies rose in 46 of 51 states. More intervention at birth; mortality at birth not improving. We are not claiming those two facts explain each other. We are saying they belong on the same page and nobody had put them there.
Two series, not one. The quarterly chart is provisional and rolling; the race chart is final and annual. We have deliberately not spliced them — joining a provisional series to a final one produces a step that looks like a finding and is an artefact.
Do not lead with the American Indian and Alaska Native number. It is the highest rate of any group in 2024 at 5.3, and it is clearly above the 3.8 of 2020 and 2021. But the series bounces — 3.8, then 5.2, then 4.8, then 5.3 — and its intervals are the widest on the chart, [4.5–6.4] in 2024 against [2.8–3.0] for White infants. The level is worth reporting. A percentage change between any two of those years is not, and we have deliberately not calculated one.
Race here is the mother’s, as recorded on the birth certificate, and single-race non-Hispanic categories. Reporting of American Indian and Alaska Native race on death certificates is known to be incomplete, which biases that rate downward, not up.
Both series are public, free, and need no key or login.
| Quarterly series | NCHS Vital Statistics Rapid Release, Quarterly provisional estimates for infant mortality. Open dataset jqwm-z2g9 on data.cdc.gov. Rates are 12-month-ending, per 1,000 live births; cause-specific rates per 100,000. Covers 2023 Q1 through 2025 Q4. |
|---|---|
| Annual series | NCHS Data Query System, Infant, neonatal, and postneonatal mortality rates by detailed race and Hispanic origin of mother. Open dataset j7ym-uwqy on data.cdc.gov. Final rates with published 95% intervals; covers 2017 through 2024. |
| Definitions | Neonatal = death before 28 completed days. Postneonatal = death from 28 through 364 days. Infant = death before the first birthday. All per 1,000 live births. |
| Cross-check | The annual final rate for all mothers in 2024 (3.7) agrees with the provisional 12-month-ending rate at 2024 Q4 (3.67), which is what you want to see from two independently produced series. |
| Pulled | 11 August 2026. Quarterly file last updated by NCHS 26 May 2026; annual file 4 June 2026. |
Cite as: Labora Rounds, Neonatal and postneonatal mortality in the United States, 2017–2025. Derived from NCHS VSRR and NCHS DQS via data.cdc.gov; pulled 11 August 2026.
Healthy, full-term, normal-weight newborns are going to intensive care at rising rates in almost every state. Same period, different measure, built from the birth records rather than the death records.