Two days before smoke policy made the news, the smoke made its own case. On July 16, 2026, 50 official EPA monitors recorded a 24-hour air quality index in the "Hazardous" range across five states, plus nine more in Ontario. Detroit was the most polluted major city in the world for most of the day, before Chicago took the title that night.

For scale, the June 2023 event that turned New York orange and became the defining image of North American wildfire smoke produced about three monitors in the hazardous range, all in Pennsylvania. Fifty against three is not an increment.

Which raises a question about a regulatory provision most people have never heard of: what happens to a rule built around the premise that this is rare?

What the Exceptional Events Rule does

Under Clean Air Act Section 319(b), the Exceptional Events Rule allows the exclusion of air quality monitoring data influenced by exceptional events from regulatory decisions regarding the National Ambient Air Quality Standards.

Exceptional events are defined as unusual or naturally occurring events that affect air quality but are not reasonably controllable using techniques that tribal, state, or local air agencies may implement to attain the standards. The category covers wildfires, high wind dust events, prescribed fires, stratospheric ozone intrusions, and volcanic and seismic activity.

The logic is sound and worth stating fairly, because the rule is not a loophole in origin. If a fire in Alberta pushes Houston's ozone above the standard, penalizing Texas accomplishes nothing. Texas cannot regulate Alberta. Air quality regulation is supposed to drive action on controllable sources, and holding a state responsible for uncontrollable ones imposes costs without producing health benefits. Texas's own filing on 2022 ozone exceedances put it exactly that way: because of the wildfires, the ozone concentrations were not reasonably controllable or preventable.

That reasoning holds cleanly when smoke is an anomaly. It holds less cleanly when smoke is a season.

The divergence

Peer-reviewed analysis of the rule reaches a conclusion that should be the center of this discussion. Increasing wildfire activity in the western US and Canada has driven declining air quality in some US regions, and because states can exempt smoke-affected readings, wildfire smoke is leading to a growing divergence between actual and regulatory air quality.

Two air qualities, drifting apart. One is what enters people's lungs. The other is what the compliance system records.

Note also that smoke is not only a particulate problem. Wildfires produce volatile organic compounds and nitrogen oxides, which combine in the atmosphere to produce ozone, and some large fires significantly affect ambient ozone readings in urban areas. So exclusions can apply to both the PM2.5 standard and the ozone standard, widening the affected footprint.

The second-order effect that matters most

Here is the consequence that gets least attention, and it is not primarily about accounting honesty.

The NAAQS system is not a scoreboard. It is a trigger. When an area exceeds a standard, it is designated nonattainment, which obligates the state to produce an implementation plan imposing additional emission controls on local sources: traffic, industry, power generation, consumer products.

Now consider a metro area whose ozone problem is, hypothetically, sixty percent local sources and forty percent smoke. Exclude the smoke-affected days and the area may fall into attainment, which switches off the requirement to further control the local sixty percent.

But a resident's lungs receive the whole hundred percent. Their asthma does not distinguish between the ozone the law counts and the ozone it excused. So the population takes a double hit: elevated total exposure, and a regulatory system that has stopped requiring reductions in the very sources that remain controllable.

That is the perverse outcome. The exclusion is justified on the grounds that smoke is uncontrollable, which is true, and its effect is to reduce pressure on the sources that are controllable, which is not what anyone intended. As smoke contributes a larger share of total exposure, this dynamic strengthens rather than fades.

The rule is getting easier to use, not harder

The administrative trend is toward more exclusions, not fewer, which is worth knowing when assessing where this goes.

EPA has released enhanced tools to help state, local, and Tribal air agencies expedite and prepare exceptional events demonstrations, including an updated Expedited Modeling of Burn Events Results tool that now supports ozone demonstrations and provides information on ozone impacts from domestic, Canadian, and Mexican fires at U.S. monitoring locations from 2021 to 2025. EPA has also produced demonstration templates for prescribed fires and tiering documents for PM2.5.

Filing an exceptional events demonstration has historically been laborious, requiring six separate elements of evidence, which meant the exemption was rationed by administrative friction rather than by policy design. Reducing that friction is defensible on its own terms, agencies should not spend months proving something obvious, and it also means the number of excluded readings will rise as the process gets cheaper. The stated rationale is that agencies can spend less time sorting wildfire smoke data and more time supporting prescribed fire work, which is a genuine benefit and also a reason to expect more exclusions.

The prescribed fire knot

There is a real policy tangle here that resists easy answers.

Prescribed burning reduces fuel loads and therefore reduces future catastrophic wildfire and the enormous smoke it produces. It is one of the few genuinely effective interventions available. It also produces smoke in the near term, and EPA's 2016 revisions recognized this by establishing that prescribed fire on wildland can be a human-caused event eligible for treatment as an exceptional event.

That accommodation is necessary. If air agencies had to count prescribed burn smoke against their compliance, they would discourage the burning, and the result would be worse fires and more smoke later. So the rule has to permit some intentional smoke to prevent much more unintentional smoke.

The difficulty is that the same exemption mechanism now covers both the fire a state deliberately set to reduce future risk and the fire that burned because nobody managed the fuel load for decades. Those are different in every way that matters, and the rule treats them similarly.

The fix that has been proposed

The researchers examining this have a specific and elegant suggestion, and it addresses precisely that distinction: modify the Exceptional Events Rule so that wildfire emissions would be excluded from air quality regulations only if states adopt government-defined best fire management policies.

Conditioning the exemption on fire management practice separates smoke a state could not have influenced from smoke it contributed to through inaction. The exemption stops being automatic and becomes earned, which converts it from a pure accounting relief into an incentive for the mitigation that actually reduces future smoke.

The same research examined the degree to which smoke causing air quality violations is driven by out-of-state fires and fires on federal lands, which is the necessary empirical groundwork: any conditional scheme has to distinguish what a state could plausibly have managed from what it could not. Smoke from federal land inside a state's borders is a particularly awkward case, since the state bears the air quality consequence without controlling the land.

What is actually at stake

The narrow question is how to score compliance. The broader one is whether the country's primary air quality measurement system will continue to reflect the air people breathe.

A regulatory framework that systematically excludes a growing share of actual exposure will report improvement while conditions worsen, and it will do so in good faith, because each individual exclusion is justified under rules written for a different world. That is how measurement systems fail: not through manipulation, but by continuing to apply a sound definition after the underlying reality has moved. Fifty hazardous monitors in one day is the signal. The one option that does not work is leaving the framework alone and treating the resulting numbers as a description of the air.

Further reading