Overdose Deaths Fell 14 Percent in 2025. The Decline Is Already Slowing.

The Centers for Disease Control and Prevention reported in May that approximately 69,973 Americans died of drug overdoses in 2025 — down from 81,313 in 2024, a decline of roughly 14 percent. It is the largest single-year drop in overdose deaths in recorded U.S. history. At the peak in 2023, more than 107,000 people died. The line is unmistakably moving in the right direction.

Sixty-nine thousand people still died.

This distinction — between the trend line and the absolute number — is the entire interpretive problem with the current moment in overdose policy. The decline is real. It is driven primarily by falling fentanyl-related fatalities, the result of a combination of factors including expanded naloxone access, more people on medication-assisted treatment, and what researchers at ScienceDirect have identified in preprint as possibly decreased average fentanyl potency in the illicit drug supply. STAT News coverage noted in January that the decline represents the longest sustained drop in decades.

What the same data shows, if you look closely, is that the rate of decline is slowing. The 12-month period ending August 2025 showed approximately 73,000 deaths — a 21 percent drop from the prior year. The full-year 2025 figure of 69,973 implies the pace is decelerating. We went from a 21 percent pace to a 14 percent pace in the space of a few months of data. The question is whether the line continues down, plateaus, or reverses.

That question is not academic. The Medicaid work requirement rule finalized on June 1 threatens to disrupt treatment access for more than 1.6 million people with SUD. The SAMHSA consolidation proposal — though rejected by Congress in the enacted FY26 budget — will likely resurface in future appropriations cycles. Harm reduction infrastructure that was built and expanded over the past decade is under regulatory pressure in several states. These are headwinds. They hit a declining trend at exactly the wrong moment.

The 14 percent drop is a reason to understand what is working and protect it. It is not a reason to assume the problem is solved. It is not a reason to cut the programs that drove it.

Sixty-nine thousand is not a success. It is progress, which is different. Progress requires active maintenance. The data on what happens when it doesn’t have that is already in the historical record: after the first wave of the opioid crisis appeared to slow in the early 2010s, the response was to declare partial victory. Then fentanyl arrived.

For historical and ongoing overdose data, see The Crisis, By the Numbers.