Fentanyl Deaths, County by County: An Interactive Map, 1999 to 2024

Fentanyl.

To put how deadly fentanyl is in context, consider that from 2013 to 2022, yearly overdose deaths in the U.S. went up by 63,959. Over the same years, deaths involving fentanyl went up by 70,733. Overdose deaths that did not involve fentanyl went down by 6,774.

Given the media attention around prescription opioids over this decade, this was surprising to me. But the data is clear: deaths involving fentanyl rose by more than all overdose deaths combined.

And then in 2024, overdose deaths in America fell sharply. Why? Fentanyl deaths fell sharply. Overdose deaths fell by 25,623, and deaths involving fentanyl fell by 25,041, or 98% of the drop. In short, since 2013, the rise and fall in drug overdoses has simply been a story of the rise and fall of fentanyl overdoses.

The chart below shows overdose deaths each year, split into deaths that involved fentanyl and deaths that did not.

Source: CDC WONDER, Multiple Cause of Death.

If understanding drug overdoses means understanding fentanyl, we need to understand fentanyl. So, today we are publishing a free, interactive map of fentanyl deaths in every U.S. county for every year from 1999 to 2024, a short history of how fentanyl came to America, a look at the counties that stand out, and what is known, and not known, about why deaths fell in 2024.

Open the full map for the title and the full notes.

How to read it

• Darker red means a higher death rate, per 100,000 residents.
• Gray counties had 0 to 9 deaths. CDC's data rules bar publishing counts that small, so the map cannot show which of these counties had none.
• Hatched counties had 10 to 19 deaths. Rates based on so few deaths can swing a lot from year to year, so we call these out for you.
• DC has its own small box, east of the coast, with a line pointing to it.

In 1999, only 6 counties had 10 or more fentanyl deaths. In 2013 there were 55. By 2022 there were 991, and those counties held 94% of the country's fentanyl deaths.

How fentanyl got here

Fentanyl started as a medicine. A Belgian chemist, Paul Janssen, first made it around 1960 (review, *International Journal of Molecular Sciences*). The FDA approved it as an injected anesthetic called Sublimaze in 1968 (CDEK; FDA) and as the Duragesic skin patch in 1990 (FDA). DEA says it is "approximately 100 times more potent than morphine and 50 times more potent than heroin" as a painkiller (DEA).

Illegally made fentanyl showed up long before the recent crisis. In 1979, fentanyl-type drugs sold as "China White" appeared on the streets in California (*Journal of Forensic Sciences*). From April 2005 to March 2007, CDC and DEA tied 1,013 deaths to illegally made fentanyl, most of them in the Chicago, Philadelphia and Detroit areas. Deaths peaked in June 2006, a month after police closed a fentanyl lab in Toluca, Mexico (CDC). DEA then put limits on two chemicals used to make fentanyl, in 2007 and 2010 (DEA). You can see part of that outbreak on the map in 2006: Cook County, Illinois, had 283 fentanyl deaths and Wayne County, Michigan, had 103.

Source: CDC WONDER, Multiple Cause of Death.

Philadelphia, where CDC counted 269 outbreak deaths, shows fewer than 10 on the map in each of those years. We have not found the reason, and it is one more sign that counts from early years should be read with care.

CDC describes the opioid crisis in three waves (CDC). The first began with more opioid prescribing in the 1990s. The second began in 2010 with a fast rise in heroin deaths. The third began in 2013, with deaths involving synthetic opioids, "particularly those involving illegally made fentanyl." On the map, the U.S. fentanyl death rate was 1.0 per 100,000 people in 2013 and 22.2 in 2022. From 2014 to 2015 alone, CDC reported that the death rate for this group rose 72%, a rise it said was "likely driven primarily by illicitly manufactured fentanyl" (CDC). The prescription opioid and heroin waves drew wide attention, but neither death rate went above 4.8 per 100,000 people. The fentanyl death rate peaked at 22.2 in 2022. That is 363% higher than the highest rate for heroin (4.8 in 2016 and 2017) and 393% higher than the highest rate for prescription opioids (4.5 in 2016 and 2017). The chart below shows the three waves in the death rate for each type of opioid. Look at the scale on the left side and compare the three lines.

Source: CDC WONDER, Multiple Cause of Death.

DEA has described how the supply worked. In 2018 it told Congress that fentanyl and the chemicals used to make it were "often produced in China" and shipped by mail to the U.S., or to groups in Mexico, Canada and the Caribbean. There it was mixed into heroin and other drugs or pressed into pills (DEA testimony, 2018). Drug labs reported finding fentanyl 36,061 times in 2016, up from 1,042 in 2013. By 2023, DEA said the Sinaloa and Jalisco cartels bought the chemicals in China, made fentanyl in Mexico, and pressed it into fake pills that look like prescription drugs. It said 6 in 10 of the fake pills it tested held a potentially deadly dose (DEA testimony, 2023). DEA counts 2 milligrams of fentanyl as a potentially deadly dose (DEA).

Fentanyl is also mixed into other drugs. DEA says it "is often mixed with other drugs, such as heroin and cocaine," and that people may not know it is there (DEA). In 2025, DEA reported that fentanyl purity was going down, while fentanyl mixed with animal tranquilizers was on the rise (DEA).

The counties that stand out

In this section we count only counties with 20 or more fentanyl deaths in a year, the same rule the map's ranking uses. Small counties can post very high rates from a handful of deaths.

Southern West Virginia has the highest rates on record. McDowell County, West Virginia, had 206.4 fentanyl deaths per 100,000 residents in 2023 (36 deaths), the highest rate for any county in any year. If you rank counties by their worst year, eight of the top ten are in West Virginia: McDowell, Wyoming, Logan, Mercer, Raleigh, Cabell, Mason and Fayette. The other two are Baltimore city and Scioto County, Ohio.

Source: CDC WONDER, Multiple Cause of Death.

Baltimore stands out among big cities. Baltimore city had the highest rate in the country in 2024, 95.2 per 100,000 (541 deaths). Its worst year was 2021, at 135.6 per 100,000 (782 deaths). Baltimore city, Cabell County, West Virginia (home to Huntington), and St. Louis city were in the country's top 25 by rate every year from 2016 to 2024.

High rates are not only in the East. In 2024, Rio Arriba County, New Mexico, had the second-highest rate in the country (80.1 per 100,000, 32 deaths). Anchorage, Alaska, was seventh (55.9 per 100,000, 162 deaths).

The most deaths are in big counties with lower rates. In 2024, Maricopa County, Arizona, had 1,005 fentanyl deaths (21.5 per 100,000), Cook County, Illinois, had 990 (19.1) and Los Angeles County had 988 (10.1).

2024: a drop almost everywhere

Fentanyl deaths fell from 72,776 in 2023 to 47,735 in 2024. The death rate fell 35.6%, more than for any other drug group CDC tracks (CDC).

Every state with numbers for both years fell. That is 49 states, counting DC. Hawaii's 2024 total and New Hampshire's 2023 total are not shown, to protect small county counts. The largest drops were in West Virginia (down 48.4%), Arkansas (46.6%), North Dakota (46.1%), Virginia (45.8%) and Wisconsin (45.7%). The smallest were in Nevada (down 4.9%), Alaska (6.9%), Utah (9.8%), Washington (13.5%) and Arizona (16.5%).

Source: CDC WONDER, Multiple Cause of Death.

Nearly every county fell too. Of 443 counties with 20 or more fentanyl deaths in both 2023 and 2024, 417 (94.1%) had fewer deaths in 2024. Six were unchanged and 20 rose. With a lower bar, 10 or more deaths in both years, 645 of 714 counties (90.3%) fell.

It was not a change in how deaths were recorded. All overdose deaths fell 24%. Overdose deaths that list an unspecified drug (CDC code T50.9) also fell, from 39,160 to 29,771, so fentanyl deaths were not just moved into an "unspecified drug" group.

The drop was almost all fentanyl. The number of fentanyl deaths fell 34.4%. Overdose deaths that did not involve fentanyl fell 1.8%, from 32,231 to 31,649.

The counties that went the other way

Only twenty counties with 20 or more fentanyl deaths in both years had more deaths in 2024 than in 2023. The chart below shows each of them.

Source: CDC WONDER, Multiple Cause of Death.

Spokane County, Washington, had the largest increase, from 176 to 249 deaths (31.9 to 44.8 per 100,000). Clark County, Nevada, home to Las Vegas, had the most deaths of the group, 519 in 2024.

What may explain the drop?

The short answer is that it is not settled. CDC named several possible reasons: wider use of naloxone (the overdose-reversal drug), better access to treatment, changes in the illegal drug supply, and prevention programs such as Overdose Data to Action. CDC did not rank them or measure how much each one did (CDC).

Researchers have tested some of these ideas:

• The drug supply. A paper in Science argues that a disruption in the fentanyl supply in 2023 was most likely the main driver (University of Maryland summary). DEA reports that 29% of the fentanyl pills it tested in fiscal year 2025 held a potentially deadly dose, down from 76% two years earlier (DEA).
• Only fentanyl deaths fell. A study of the same CDC data we use found the same split we did, with fentanyl deaths falling sharply and other overdose deaths barely moving. The author says this fits a change in the fentanyl supply better than broad program effects. That study has not been peer-reviewed (medRxiv).
• Drug seizures. A study in JAMA Health Forum tied about 9% of the drop to fentanyl's falling share of drug seizures, and its authors say other factors are likely at work (study).
• Naloxone. A study in JAMA found that emergency crews gave naloxone at about the same rate per overdose as before the drop. Nonfatal opioid overdoses seen by emergency crews were 32.7% below what earlier trends predicted. The authors say this suggests fewer overdoses were happening, not that more people were being revived. The study did not count overdoses reversed without a 911 call (Brown University).
• Pushback on the supply idea. A peer-reviewed paper in Science Advances found that national fentanyl purity "did not track fatal fentanyl overdose rates in most United States regions" and does not explain the drop on its own (USC Schaeffer).

In our view, the evidence leans most strongly toward changes in the drug supply (for example, less fentanyl in each pill). We could not find a study that measures how much treatment, naloxone or prevention programs added.

Looking ahead

What worries us is the unknown. Deaths fell fast in 2024, but the people who study it do not agree on why. If the drop came mostly from changes in the drug supply, it could reverse if the supply changes again, resulting in many more deaths. The authors of the Science paper write that "How long the decline will last is uncertain" and that "Treatment, prevention and harm-reduction efforts remain essential" (University of Maryland summary). The supply is still changing. DEA warns that medetomidine, "a more powerful veterinary anesthetic," has started to show up in fentanyl (DEA). And deaths are still high: CDC's early count is 67,798 overdose deaths in the 12 months ending in March 2026 (CDC).

That is why the funding matters. Cutting programs on the belief that the problem is solved would remove support while researchers still disagree about what is holding deaths down.

That said, let's look at funding.

What Congress has done. For fiscal year 2026, Congress kept funding of the main programs stable. The State Opioid Response grants, which pay for treatment, naloxone and recovery services in every state, are $1.595 billion. The substance use block grant to states is $2.013 billion (GHAI). Funding for fiscal year 2027 has not passed. For now, programs run at 2026 levels through December 11, 2026 (Congressional Research Service).

What has been proposed. HHS's budget request for fiscal year 2027 would (HHS):

• fold SAMHSA, the agency that runs most of these grants, into a new agency called the Administration for a Healthy America. HHS also proposed this agency for fiscal year 2026, and Congress did not create it (GHAI)
• merge the State Opioid Response grants ($1.595 billion), the substance use block grant ($2.013 billion) and the mental health block grant ($1.012 billion) into one $4.621 billion block grant, the same total as today (HHS)
• cut SAMHSA's other substance abuse prevention programs from $350 million to $89 million, and its other treatment programs from $564 million to $490 million. These are separate from the State Opioid Response grants.
• move CDC's injury center, which runs its overdose programs, into that new agency at $588 million, down from $761 million

These are proposals. Congress decides the final amounts.

Grants cut and then restored. In January 2026, SAMHSA ended about $2 billion in grants and reversed the decision the next evening (NACo). In mid-2025, about $140 million in Overdose Data to Action money was held back, and CDC staff later said they expected it to be paid in full (KAXE/NPR).

Medicaid. This may be the biggest concern, in my view. Medicaid covers about one-fifth of adults with a substance use disorder (KFF). The 2025 budget law adds work requirements and eligibility checks every six months, starting by December 31, 2026, though some states may get until the end of 2028. While people in substance use treatment are exempt from the work requirements (KFF), talks I have had in my community have made me very concerned that staffing at the state and county offices that handle Medicaid eligibility may not be able to keep up with the new, more complex eligibility checks, leading to people who should still qualify for Medicaid not receiving it.

Two earlier cases point the same way:

• Arkansas, 2018. Arkansas was the first state to enforce Medicaid work requirements, and 18,000 adults lost coverage in a few months. A study in the New England Journal of Medicine found that more than 95% of the people subject to the rule already met it or should have been exempt (Harvard).
• The end of pandemic coverage, 2023 to 2024. When states restarted yearly eligibility checks, at least 25.2 million people were dropped from Medicaid. Of those, 69% lost coverage for procedural reasons, such as missing paperwork, rather than being found ineligible (KFF).

We plan to investigate this further as it won't just impact people living with a substance use disorder, but all people who rely on Medicaid for health care and social assistance.

How we built it

Every number on the map can be checked and rebuilt. The full method is in DATASETS.md, and the checks are in TIEOUT.md.

• Source. CDC WONDER, Multiple Cause of Death: the 1999-2020 database for 1999 to 2017, and the 2018-2024 Single Race database for 2018 to 2024. The two databases give the same county counts for 2018 to 2020.
• What counts as a fentanyl death. A death whose underlying cause was a drug overdose (accidental, suicide, homicide or undetermined) and whose death certificate lists synthetic opioids other than methadone (ICD-10 code T40.4).
• Where. Deaths are counted in the county where the person lived, not where they died.
• Rates. Crude rates per 100,000 residents in every year, because CDC does not give county age-adjusted rates in its newer database. For 1999 to 2017, CDC does not show rates for counties with 10 to 19 deaths, so we calculate them the way CDC does: deaths divided by population, times 100,000.
• Small counts. CDC does not publish county counts of 1 to 9, and its data rules bar publishing any count of 9 or fewer. Those counties are gray. Where a state's total would let a reader work out a small county count by subtraction, we leave the state's total out for that year (43 state-years).
• Checks. A script recomputes every number on the map from CDC's files, about 250,000 values, with no differences. A separate check, written from scratch, pulled the data again and found no differences.

Reproduce it yourself

The code, the data and the full method are at github.com/Data4ThePeople/FentanylDeaths. If you get a different number from us, tell us, and we will look.