Why Has Ozone Stopped Improving?
EPA monitor and emissions data through 2024–2025 · Updated August 13, 2026 · Methods
Not because the country stopped cutting pollution. Emissions of nitrogen oxides (the main ingredient of ozone) fell 44% in the eastern United States between 2014 and 2024, nearly the same rate as the 53% cut in the twelve years before. Measured power-plant emissions fell 60%. Ambient ozone stopped responding. The eastern fourth-highest reading fell 26 ppb between 2002 and 2014, then essentially flat-lined. The obvious explanation, that the emission cuts ran out, is wrong, and the data says so plainly.
Emissions and ambient levels are not the same measurement. Emissions are tons released, measured at the stack by continuous monitors or estimated for vehicles. Ambient levels are what a monitor detects in the air, in parts per billion. Ozone is not emitted at all: it forms when nitrogen oxides and volatile organic compounds react in sunlight. Everything on this page turns on the difference: the inputs kept falling and the output stopped following.
Emissions: still falling
| Eastern NOx emissions | 2002–2014 | 2014–2024 |
|---|---|---|
| All sources | -53% | -44% |
| Power plants (measured at the stack) | -64% | -60% |
| On-road vehicles | −63% | -65% |
Power-plant emissions are the reliable number here: they are measured by continuous monitors in the stack, not modelled. On-road vehicle emissions come from a model, so their trend carries more uncertainty. But the measured series shows the same pattern, so the conclusion does not depend on the modelled part.
Ambient ozone: stopped following
The same emission cuts that bought 26 ppb of cleaner air in the first period bought essentially nothing in the second. Comparing states to each other makes it sharper: in 2002–2014 each 1% cut in a state's nitrogen oxide emissions came with about 0.50 ppb of ozone improvement. In 2014–2024 that fell to 0.11 ppb, and the range of uncertainty now includes zero. A 52% emissions cut bought 15 ppb in the first decade; a 45% cut bought 2.5 ppb in the second.
And the West is a different problem entirely
Splitting the country apart resolves what a national average hides. In the East, the population living above the standard is still shrinking, by about 1.3 million a year. In the West it is not shrinking at all. The western share of everyone breathing above-standard air has climbed from 26% in 2002 to 42% today.
At high elevation in the interior West, the cleanest days now sit at about 39 ppb, against 24 ppb in the East. Those sites intercept air from high in the atmosphere carrying ozone from outside the region. Where a summer's quietest days already start at 39 ppb and the standard is 70, domestic emission cuts have far less room to work. Western ozone also tracks wildfire years closely, though fire smoke and the hot dry weather that causes fires cannot be told apart in this data, because both come from the same summers.
What this means, and what it does not
“The East ran out of emission reductions” would imply the remedy is more regulatory effort. That is not what happened: the cuts continued at full pace and bought very little ambient improvement. The binding constraint has moved from how much pollution is released to how the atmosphere converts it, which is a question about chemistry, not ambition.
We cannot say why. Several explanations fit: ozone formation is not proportional to nitrogen oxides and can saturate; some urban areas respond to volatile organic compounds rather than nitrogen oxides; background ozone arriving from outside the country may be rising; warmer summers favour ozone formation and could be offsetting emission cuts. Distinguishing these needs satellite chemistry measurements or an atmospheric model, and this analysis cannot separate them. What it establishes is that the emissions explanation is ruled out.
Sources
- EPA Air Pollutant Emissions Trends, state by sector, 2002–2024. Power-plant figures derive from continuous stack monitoring; on-road figures from EPA's MOVES model.
- EPA AirData daily ozone monitor files, 2000–2025, warm season (May–September).
- Monitor elevations from EPA's AQS site metadata; regional split at 100°W and 1,500 m.
- Methods and validation.