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 U.S. nitrogen oxide emissions by year, million short tons 0.0 5.0 10.0 15.0 20.0 2002 2006 2011 2015 2020 2024 2002: 20.8M tons 2003: 20.2M tons 2004: 18.6M tons 2005: 17.7M tons 2006: 16.4M tons 2007: 15.3M tons 2008: 14.5M tons 2009: 12.5M tons 2010: 12.3M tons 2011: 11.8M tons 2012: 11.0M tons 2013: 10.3M tons 2014: 9.8M tons 2015: 8.8M tons 2016: 7.9M tons 2017: 7.4M tons 2018: 7.1M tons 2019: 6.8M tons 2020: 5.9M tons 2021: 6.0M tons 2022: 5.9M tons 2023: 5.6M tons 2024: 5.5M tons
Eastern U.S. nitrogen oxide emissions, million short tons a year, all sources. EPA National Emissions Inventory.
Eastern NOx emissions2002–20142014–2024
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

Eastern U.S. warm-season ozone, fourth-highest daily maximum, ppb 0 20 40 60 80 2000 2005 2010 2015 2020 2025 2000: 81 ppb 2001: 82 ppb 2002: 89 ppb 2003: 78 ppb 2004: 71 ppb 2005: 79 ppb 2006: 76 ppb 2007: 78 ppb 2008: 70 ppb 2009: 65 ppb 2010: 69 ppb 2011: 72 ppb 2012: 75 ppb 2013: 64 ppb 2014: 63 ppb 2015: 65 ppb 2016: 65 ppb 2017: 64 ppb 2018: 66 ppb 2019: 61 ppb 2020: 60 ppb 2021: 63 ppb 2022: 62 ppb 2023: 68 ppb 2024: 63 ppb 2025: 63 ppb 2015 standard, 70 ppb
Eastern U.S. warm-season ozone, median across monitors of the fourth-highest daily maximum 8-hour reading, the statistic the standard is judged on. Parts per billion.

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.

Intermountain West high-elevation ozone, 5th percentile, ppb 0 10 20 30 40 2000 2005 2010 2015 2020 2025 2000: 34 ppb 2001: 34 ppb 2002: 35 ppb 2003: 38 ppb 2004: 32 ppb 2005: 34 ppb 2006: 37 ppb 2007: 37 ppb 2008: 35 ppb 2009: 36 ppb 2010: 36 ppb 2011: 40 ppb 2012: 41 ppb 2013: 38 ppb 2014: 37 ppb 2015: 36 ppb 2016: 36 ppb 2017: 37 ppb 2018: 40 ppb 2019: 38 ppb 2020: 38 ppb 2021: 41 ppb 2022: 39 ppb 2023: 40 ppb 2024: 41 ppb 2025: 39 ppb
The floor is rising in the Intermountain West: the 5th percentile of warm-season ozone at monitors above 1,500 metres, parts per billion. Even the region's cleanest days are getting worse.

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