Editorial correction — September 20, 2026. We retain the dated provisional finding, remove an unsupported war-death comparison and speculative causal explanation, and clarify that a national decline can coexist with increases in particular states. Expanded September 20, 2026 with additional primary-source data and a Sources list.
Fewer Americans died of drug overdoses in 2025 than in 2024, according to CDC’s provisional estimate, and 2024 was already far below the 2022 peak. That is worth saying plainly. It also comes with limits: the 2025 number is provisional, the decline was not shared equally, and roughly 70,000 deaths is still a national emergency by any ordinary standard. This article sets out what the federal data show, how to read them, and what remains unfinished.
What the 2025 estimate says
On May 13, 2026, CDC’s National Center for Health Statistics (NCHS) released provisional data on overdose deaths. It estimated 69,973 deaths in 2025, down almost 14% from an estimated 81,313 in 2024, and described this as the third consecutive year of decline. Opioid-involved deaths fell from 55,296 to 44,564 over the same comparison, a drop of about 19% by simple arithmetic on the release’s own figures. The release also reported that deaths involving cocaine and psychostimulants such as methamphetamine decreased.
Almost every state saw a decline. Five states (Rhode Island, New York, North Carolina, Alabama and Vermont) recorded drops of 25% or more. Three (New Mexico, Arizona and Colorado) recorded increases of at least 10%. A falling national total therefore did not mean improvement everywhere.
Why provisional numbers move
Overdose deaths take time to count. Investigators and toxicology labs have to finish their work before a death can be classified. NCHS publishes two kinds of figures: reported counts, which are the deaths received and processed so far, and predicted counts, which are adjusted for incomplete reporting. NCHS’s own documentation warns that provisional counts can substantially understate final counts where many deaths are still pending investigation, potentially by as much as 30% in some jurisdictions. That is why the 2025 figure is an estimate and can be revised.
You can see the difference between series in the 2024 numbers. The May 2026 release put the 2024 provisional estimate at 81,313. The final NCHS count for calendar year 2024 is 79,384. Both figures are legitimate. They come from different products at different stages. The reading rule is simple: compare numbers within one series, and never subtract a provisional figure from a final one.
The long view: from a peak to a lower level
The final NCHS series shows how large the change has been. The age-adjusted death rate, which removes the effect of a changing age mix so that years can be compared, rose from 8.9 per 100,000 people in 2003 to 32.6 in 2022, the peak. It fell 4.0% in 2023 and 26.2% in 2024.
| Year (final data) | Overdose deaths | Age-adjusted rate per 100,000 |
|---|---|---|
| 2022 | 107,941 | 32.6 |
| 2023 | 105,007 | 31.3 |
| 2024 | 79,384 | 23.1 |
Source: NCHS data briefs. The provisional 2025 estimate (69,973) belongs to a different series and is not included in this table.
Which drugs, and which people
In 2023, 72,776 of the 105,007 overdose deaths, or 69.3%, involved synthetic opioids other than methadone, a category that includes fentanyl. That category also showed the sharpest fall in 2024.
| Drug category (rate per 100,000) | 2023 | 2024 | Change |
|---|---|---|---|
| Synthetic opioids other than methadone | 22.2 | 14.3 | -35.6% |
| Psychostimulants with abuse potential (mainly methamphetamine) | 10.6 | 8.5 | -19.8% |
| Cocaine | 8.6 | 6.3 | -26.7% |
| Natural and semisynthetic opioids | 2.9 | 2.3 | -20.7% |
| Heroin | 1.2 | 0.8 | -33.3% |
| Methadone | 1.0 | 0.9 | -10.0% |
A death can involve more than one drug, so these categories overlap and should not be added together. Changes for natural and semisynthetic opioids, heroin and methadone are calculated here from the rounded rates shown, so treat them as approximate.
The 2024 data also show who was affected. Men accounted for 55,076 deaths (32.2 per 100,000) and women for 24,308 (14.1). By age, the highest rate was among people 35 to 44 (44.2 per 100,000), followed by ages 45 to 54 (41.0) and 55 to 64 (38.6). The rate for ages 15 to 24 was 8.5, down 37.0% from 2023.
Where the decline did not reach
The improvement was uneven. Adults 65 and older had a rate of 13.4 per 100,000 in 2024, and their decline of 8.8% was the smallest of any age group in the NCHS brief. By race and ethnicity, American Indian and Alaska Native people had the highest rate at 51.6 per 100,000, compared with 24.7 for white non-Hispanic people, 20.5 for Native Hawaiian and Pacific Islander people, 17.0 for Hispanic people and 4.4 for Asian non-Hispanic people. Black non-Hispanic people had a rate of 33.8 after a 30.9% decline from 2023.
Add the state picture from the May 2026 release and the point is clear. A national average is not a local service directory. A community’s own numbers can move in a different direction from the country’s.
Treatment: the distance between need and care
Deaths are one measure. Access to care is another. The Substance Abuse and Mental Health Services Administration’s 2024 National Survey on Drug Use and Health estimated that 4.8 million people aged 12 or older had an opioid use disorder in the past year. It also estimated that 52.6 million people needed substance use treatment, including 48.4 million with a substance use disorder, and that 10.2 million of them, or 19.3%, received treatment.
The share who received treatment varied by age: 11.3% of young adults 18 to 25 who needed it, 20.5% of adults 26 and older and 30.2% of adolescents 12 to 17. Separately, the survey estimated that 2.2 million people, 0.7% of the population aged 12 and older, received medication for opioid use disorder. Do not divide 2.2 million by 4.8 million. The two figures use different bases and classifications, so they are not a numerator and a denominator.
A treatment-gap statistic is not a verdict on any single program. It does describe a distance between the number of people who needed help and the number who received it, and that distance is where local decisions about funding, staffing and access do their work.
What this article does not claim
It does not say why deaths fell. The federal releases report what was counted, not the cause, and no single explanation can be read off an annual total. Treatment, prevention, changes in the drug supply and shifts in who remains at risk may all matter. These numbers cannot rank them, and anyone who claims otherwise from a single chart is claiming more than the data can carry.
How to read the next release
- Check the publication date and the period the data cover.
- Check whether the figure is provisional, predicted or final, and keep that label attached.
- Look for state and drug detail before generalizing from the national number.
- Compare like with like: final to final, provisional to provisional.
- Ask what a person in your community can actually reach: treatment, naloxone, and a way to ask for help.
Progress and duty together
Fewer people dying is good news, and a publication concerned with social harm should be able to say so directly. Treating every improvement as one more route to a predetermined story of decline would be a failure to follow the evidence. At the same time, an estimate near 70,000 deaths is an enormous continuing loss. Acknowledging progress and caring about the people still harmed reinforce one another.
Sustained support should not depend on whether a crisis currently attracts attention. Families need institutions that stay dependable through alarming and encouraging news cycles alike, and people seeking help should not be reduced to statistics used to win an argument.
Sources
- CDC, National Center for Health Statistics, press release on provisional 2025 overdose deaths, May 13, 2026: cdc.gov/nchs/pressroom/releases/20260513.html
- NCHS Data Brief No. 549 (final 2024 overdose death data): cdc.gov/nchs/products/databriefs/db549.htm
- NCHS Data Brief No. 522 (2023 overdose death data and trends since 2003): cdc.gov/nchs/products/databriefs/db522.htm
- NCHS, Vital Statistics Rapid Release, Provisional Drug Overdose Data (definitions of reported and predicted counts and reporting lag): cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm
- SAMHSA, “Key Substance Use and Mental Health Indicators in the United States: Results from the 2024 National Survey on Drug Use and Health”: samhsa.gov, 2024 NSDUH annual national report