For most of the past half century, the story of American air quality was one of the clearest environmental success stories anywhere. The Clean Air Act, signed in 1970 and strengthened repeatedly afterwards, drove down the pollutants that had left industrial cities visibly hazy. Fine particulate matter — the PM2.5 that lodges deepest in the lungs — fell by roughly 40% nationally between 2000 and 2022 as coal plants retired, catalytic converters spread and diesel rules tightened. Then, somewhere around 2016, the line flattened. In much of the country it has since bent back the wrong way. The cause is not tailpipes or smokestacks. It is smoke.
The progress being undone
The clearest accounting comes from a 2023 Nature study led by Stanford's Marshall Burke with colleagues at Harvard and the University of Washington, which combined satellite and ground-sensor data from 2000 to 2022 to separate smoke from everything else. Their finding: since at least 2016, wildfire smoke has influenced annual average PM2.5 trends in nearly three-quarters of states in the contiguous United States, eroding about 25% of the previous multi-decadal progress in those states — the equivalent of roughly four years of air-quality gains wiped out. In many western states the erosion exceeds 50%.
That is the part worth sitting with. Decades of regulation and capital spending bought a real public-health improvement, and one unregulated source has clawed back a quarter of it in under a decade. Burke's own summary is blunt: the country is eroding decades of progress in the span of five to ten years, due to just this one source.
Why smoke PM2.5 is not just “bad air”
Regulators have historically treated a microgram of PM2.5 as a microgram, wherever it came from. The research increasingly suggests smoke is worse than that accounting implies. Wildfire PM2.5 contains a higher share of carbonaceous and polar organic compounds than industrial or traffic particulate, and those components appear to be more biologically aggressive. The particles act like sponges, binding toxic metals and organic compounds and carrying them into lung tissue and the bloodstream, where they drive inflammation well beyond the respiratory system.
Composition also depends on what is burning. When a fire runs from forest into a town, the plume stops being vegetation smoke. Researchers at Harvard's T.H. Chan School of Public Health, studying firefighters and civilians after smoke exposure, detected mercury, cadmium, carcinogenic compounds and PFAS — the “forever chemicals” used in building materials. As the National Institutes of Health's Aubrey Miller has put it, the question is what is burning: plastics, heavy metals, arsenic. A wildland-urban interface plume is simply a different pollutant than a remote forest plume.
The health evidence has firmed up considerably. A causal-modelling study published in Science Advances in February 2026 by Min Zhang and colleagues at Mount Sinai attributed roughly 24,100 all-cause deaths per year in the contiguous United States to chronic wildfire smoke PM2.5 exposure over 2006–2020. Two findings stand out: the exposure–response curve rose monotonically with no evidence of a safe threshold, and of the causes examined, mortality from neurological disease showed the steepest increase per unit of exposure. Separate work in the European Heart Journal links chronic smoke exposure to elevated stroke risk in older adults. Stanford researchers have found a measurable increase in preterm birth risk from a single week of elevated smoke — even during periods that federal indices rated “good” or “moderate.”
The exposure map has gone national
Smoke used to be filed under “western problem, summer season.” That framing is obsolete. Fine particles can stay suspended for around ten days absent rain, long enough to cross a continent, which is why smoke from Canadian fires has repeatedly pushed air quality indices in eastern US cities into ranges those cities had not seen in the modern measurement era. Stanford's analyses found that the number of people exposed to smoke-caused unhealthy air at least one day a year grew 27-fold in a decade, and that more than 1.5 million people now live in places routinely hit by extremely heavy smoke days.
The near-term consequences show up in emergency rooms: in the week following an extreme smoke day, visits for asthma, COPD and persistent coughing rise by 30–110%. Meanwhile the structural driver keeps strengthening. Annual acreage burned in the United States is now more than double what it was three decades ago — a product of hotter, drier fuels plus a century of fire suppression — and the United Nations Environment Programme projects the frequency of extreme fires could rise by up to 50% by 2100.
A regulatory blind spot
Here is the awkward structural fact: wildfire-driven increases in ambient PM2.5 are largely unregulated. The Clean Air Act was designed for identifiable, controllable sources — a plant you can require to install scrubbers, a fleet you can require to meet a standard. Under the EPA's Exceptional Events Rule, states and local agencies can request that wildfire-influenced exceedances be excluded from regulatory determinations. The logic is defensible in isolation: a district cannot control a fire five states away.
The problem is aggregation. If every jurisdiction can point upwind, no one is accountable for the fastest-growing source of particulate pollution in much of the country — and “exceptional” is a strange label for something that now happens every summer. Researchers have proposed a partial fix with an uncomfortable shape: exempt prescribed burns from counting against local air-quality goals, because the current rules discourage the controlled fire that reduces catastrophic fire later. Preliminary Stanford work suggests a decade-long prescribed-fire programme could cut cumulative smoke PM2.5 by roughly 23% relative to no treatment. That means accepting some smoke now to avoid much more smoke later — politically difficult, and hard to explain to a neighbourhood being asked to breathe it.
What actually reduces exposure
At the household level the evidence is unglamorous and effective. Sensor studies show that closing windows, sealing obvious gaps and running a HEPA air cleaner or a well-filtered HVAC system meaningfully lowers indoor concentrations during a smoke episode. The EPA's indoor air quality guidance for wildfires recommends designating one “clean air room” rather than filtering an entire house — cheaper, and more effective. The other high-value habit is simply checking the Air Quality Index before any outdoor exertion (thresholds in the FAQ below).
At the community level, the interventions are institutional rather than technical: clean-air shelters in libraries, schools and community centres; filtration upgrades in schools and care homes, where the most vulnerable people spend their days; and smoke-specific public health messaging instead of a generic “bad air day” advisory. Agencies are also beginning to protect the most exposed group of all. A Congressional Research Service brief published in August 2026 notes that the US Forest Service and the Department of the Interior's wildland fire service updated policies in 2026 specifically to reduce smoke exposure for fire personnel, who routinely breathe carcinogens as a condition of the job.
What it means
Three implications follow. First, air quality has become a climate-adaptation problem, not purely an emissions-control problem: you cannot put a scrubber on a forest, so the response runs through land management, building filtration and public health. Second, the “no safe threshold” finding undercuts the mental model most people use — waiting for the sky to turn orange means routinely absorbing exposure the research says still carries risk. Third, and most consequentially, the accounting is broken: as long as smoke is treated as an exceptional event, the largest driver of PM2.5 trends in much of the country stays outside the framework built to manage exactly this harm. Stanford researchers projected in a 2025 Nature paper that smoke PM2.5 could contribute on the order of 1.9 million excess US deaths between 2026 and 2055 on current trajectories. That is a large number to leave sitting in a regulatory gap.
