CDC Finds Cannabis in 1 of Every 5 Overdose Deaths — and Says the Real Story Is What It’s Riding With

Four scientists in the CDC’s Division of Overdose Prevention spent four and a half years of death-scene toxicology reports — 209,166 of them, from 31 states and Washington, D.C. — and came back with a number that will get clipped out of context by the end of the week: cannabis showed up in 21.0% of overdose deaths. Lauren Tanz, Zerleen Quader, Christine Mattson, and Alana Vivolo-Kantor published the findings September 3 in the CDC’s Morbidity and Mortality Weekly Report, and the number that should actually anchor every headline sits two paragraphs down: cannabis was the documented cause of death in just 0.8% of those cases, and the sole substance detected in nine deaths out of 209,166.

Cannabis is turning up at more death scenes, and it’s still not the thing killing people — it’s the marker of who’s already deep in a much more dangerous supply.

Translate the method first, because it matters for how much weight this number can carry. The CDC pulled its data from SUDORS — the State Unintentional Drug Overdose Reporting System, a network that combines death certificates, medical examiner and coroner narratives, and postmortem toxicology screens. A “detection” means a lab found cannabis metabolites — THC, CBD, or THCV — in a person’s system after they died of an overdose. It does not mean cannabis caused that overdose, and the CDC authors say so explicitly, writing that detection “is primarily indicative of polysubstance use” rather than a cannabis-caused fatality. Only 31 states plus D.C. feed into SUDORS, and not every jurisdiction runs the same toxicology panel, so this is surveillance data with real geographic and lab-method gaps — a real, useful signal, not a national census.

The Age Pattern Is the Actual Finding

Bury the topline number and the report’s real news is an age curve that runs backward from what you’d guess. Detection was highest among 12- to 17-year-olds at 42.7%, and lowest among adults 50 and older at 16.7% — despite older adults reporting far higher rates of any cannabis use in national surveys. That’s not because teenagers are smoking more weed than everyone assumes; a 2026 peer-reviewed scoping review on cannabis potency notes adolescent cannabis use has actually trended flat-to-down in recent years even as the products got stronger. It’s because the teenagers who die of overdoses are disproportionately teenagers who are already using multiple drugs, and cannabis is one of the things a toxicology screen happens to catch. The CDC’s own polysubstance overdose page makes the same point for every drug class: the deadliest overdoses are almost never one-drug events anymore.

The numbers that prove it sit right in the MMWR table. Among the 43,880 deaths with cannabis detected, 73.5% also involved illicitly manufactured fentanyl and 60.4% involved a stimulant. Among adolescents specifically, those co-detection rates climbed to 88.6% for fentanyl and 69.0% for stimulants. A 15-year-old who dies with cannabis in their system, in other words, is far more likely to have died with fentanyl in their system too — and this tracks with a separate 2025 Pediatrics study by Connor Buchholz and colleagues, which found that just over half of opioid overdose deaths among Americans 15 to 24 involved more than one substance class, with the share climbing steadily from age 15 to age 24. Two independent CDC-adjacent datasets, one specific to cannabis and one to youth opioid deaths, are describing the same phenomenon from different angles: young people aren’t overdosing on weed. They’re overdosing on fentanyl that showed up somewhere it wasn’t supposed to be, in bodies that also had cannabis in them.

What This Does and Doesn’t Tell You

If you’re a parent reading this because your kid uses cannabis, the finding is not “cannabis is now a leading killer of teenagers.” The finding is closer to “cannabis use in a teenager who’s also encountering an unregulated drug supply is a red flag worth taking seriously” — and the CDC frames its own recommendation the same restrained way, pointing toward community prevention programs like the Drug Free Communities Support Program rather than cannabis-specific interdiction. If you’re someone who uses cannabis yourself, alone, from a source you trust, this report says almost nothing about your risk; the danger this data is actually tracking lives in counterfeit pills and contaminated stimulants, not in a joint. The CDC’s own count of overdose deaths where cannabis was the only substance detected was nine people, nationally, across four and a half years.

The CDC’s own count of overdose deaths where cannabis was the only substance detected was nine people, nationally, across four and a half years.

The broader trend line gives some room to breathe here too: national overdose deaths have been falling for three straight years, dropping sharply enough by mid-2026 that researchers are now debating how much credit belongs to naloxone distribution versus a shifting drug supply. Cannabis detection held roughly flat against that backdrop — which is itself informative. It means cannabis isn’t driving the crisis up or down. It’s riding shotgun in an epidemic that’s actually about something else, and until toxicology panels get better at separating “present” from “responsible,” a positive cannabis screen will keep getting read as louder evidence than four CDC scientists ever claimed it was.