Now, her panel faced two more major air-pollution reviews: ozone and lead. She hoped things would get better.

They got a lot worse.

Scientists know that ground-level ozone, the chief component of smog, kills thousands of people a year. Even small changes in ozone concentrations, just 10 parts per billion over the short term, can cause death rates to climb by 0.5 percent, particularly among the elderly.

In Philadelphia alone, the EPA estimates, dozens die each year.

The noxious gas is formed by a chemical reaction when organic compounds and nitrogen oxides are exposed to sunlight. Those pollutants come from cars, power plants and oil refineries, among other sources.

It also aggravates asthma and inflames the lungs - "like a sunburn on the skin," the EPA says - and on those days when levels are high, officials warn the vulnerable to stay indoors. When levels spike, so do hospital emergency room visits.

The Clean Air Act requires the EPA to review ozone limits every five years, but the agency had not examined them since 1997, when it was set at 80 parts per billion. Ozone levels in the air are measured by about 1,200 ultraviolet monitors positioned across the United States.

In March 2003, the American Lung Association and a cadre of environmental and public-health groups sued the EPA to force a review, and a judge ordered EPA to complete one by March 12, 2008.

The law required Johnson to set two standards, a primary standard to protect human health and a secondary standard that would protect crops, plants and ecosystems.

The science panel and EPA spent almost four years assessing the risk from ozone. They all agreed that the current standard of 80 parts per billion was too high.

But what level of ozone was safe?

The panel combed through 1,700 new scientific studies to see if they could determine how much ozone, mixed into the everyday cocktail of pollution, made people sick.

These types of studies can establish relationships between pollutants and health problems, but it is easier to pinpoint a specific culprit in a controlled laboratory study.

For that, the panel turned to William C. Adams, an exercise physiologist at the University of California at Davis. Adams had conducted two studies in which healthy adults directly inhaled a controlled amount of ozone at doses near the level that was being considered for the new safety standard.

Both studies were paid for by the American Petroleum Institute.

The studies showed a statistically significant effect when subjects inhaled ozone at the current level set by the EPA, 80 parts per billion. But some adults also showed a decrease in lung function when they inhaled ozone at 60 parts per billion.

If healthy young adults found it harder to breathe with ozone at 60 parts per billion, the panel reasoned, then clearly more vulnerable populations were at risk at that level.

Adams later said he thought the panel placed too much weight on some of his findings, but the experts considered his arguments and disagreed.

In October 2006, the panel recommended that Johnson set the primary level between 60 and 70 parts per billion.

It also said Johnson should set a separate secondary standard, designed to address ozone's effects on a plant's ability to produce and store food. This secondary standard would limit cumulative ozone levels during daylight hours over three months. The panel recommended that the total not be higher than 15 "parts per million-hours" - an environmental measure akin to man-hours.