Death came to Donora, a small steel town in western Pennsylvania, in the form of a black fog.

Trapped by unusual weather conditions in October 1948, a blanket of smokestack pollution killed 20 people and sickened thousands. Overwhelmed doctors scurried to fashion oxygen tents out of spare bedsheets. The mix of deadly pollutants was so thick that when the undertaker drove to pick up one of the dead, he simply got lost.

It was the nation's most dramatic evidence of the dangers of air pollution, and it spawned a cleanup effort so successful that, today, many people rarely give a thought to the air they breathe.

Yet in June 2005, a panel of scientists appointed by the Environmental Protection Agency determined that the air was still too dirty.

The evidence was more subtle, but according to the latest studies, tiny particles - similar to those that had fallen on Donora - could shorten lifespans even at the invisible everyday levels found in much of the United States. No matter where the mix of particles came from - trucks, buses, power plants - their small size meant they could penetrate deep into the lungs.

The panel recommended tougher rules to limit long-term exposure, a move that EPA's own scientists said could prevent thousands of premature deaths annually.

EPA Administrator Stephen L. Johnson rejected their advice.

His decision took the panel by surprise, but before long, it would fit into a familiar pattern.

Over the next three years, leading environmental scientists would denounce Johnson for substituting politics for science on key pollution issues - from not regulating greenhouse gases blamed for global warming to delaying the assessment of toxic chemicals.

But it was in a succession of decisions on air quality that Johnson's uneven application of science had perhaps the most severe impacts on human health.

In the case of particles, a key element of Johnson's scientific justification was characterized by a top expert as "silly."

When considering restrictions on ozone, the chief component of smog, Johnson flatly rejected the recommendations of the nation's top 23 experts.

For a new rule on airborne lead pollution, Johnson attempted to cripple the very panel created by Congress to advise him on all three decisions.

In all three cases, it took a court order to get the EPA to make a decision.

And in all three cases, according to documents and interviews, Johnson chose to take a more industry-friendly course after pressure from the White House.

"What good is a scientific committee if you're just going to march to the White House and ask the president?" asked toxicologist Rogene Henderson, who was appointed by Johnson and chaired the panel during all three agency decisions.

Johnson, a career EPA scientist before taking the top spot in 2005, noted that the law gives him the final call on matters of science policy, and he said criticisms of him were unfair. While science is one of the "foundational principles" of a regulatory decision, he said, it's not the only one.

"For those who like the decision," Johnson said, "it's a great decision, and those who don't, you know, there's the 'you-caved-in.' "

The chorus of "you-caved-ins" has grown among scientists and other critics who say that Johnson has allowed the White House Office of Management and Budget (OMB) to unduly influence him - and that he weighs economic concerns more heavily than public safety.

Johnson himself asked Congress this year to rewrite the Clean Air Act to allow him to take into account economic impacts when setting air pollution standards.