In 2005, when government scientists tested 60 soft, vinyl lunch boxes, they found that one in five contained amounts of lead that medical experts consider unsafe _ and several had more than 10 times hazardous levels.

But that's not what they told the public.

Instead, the Consumer Product Safety Commission released a statement that they found "no instances of hazardous levels." And they refused to release their actual test results, citing regulations that protect manufacturers from having their information released to the public.

That data was not made public until The Associated Press received a box of about 1,500 pages of lab reports, in-house e-mails and other records in response to a Freedom of Information Act request filed a year ago.

The documents describe two types of tests. One involves cutting a chunk of vinyl off the bag, dissolving it and then analyzing how much lead is in the solution; the second test involves swiping the surface of a bag and then determining how much lead has rubbed off.

The results of the first type of test, looking for the actual lead content of the vinyl, showed that 20 percent of the bags had more than 600 parts per million of lead _ the federal safe level for paint and other products. The highest level was 9,600 ppm, more than 16 times the federal standard.

But the CPSC did not use those results.

"When it comes to a lunch box, it's carried. The food that you put in the lunch box may have an outer wrapping, a baggie, so there isn't direct exposure. The direct exposure would be if kids were putting their lunch boxes in their mouth, which isn't a common way for children to interact with their lunch box," said CPSC spokeswoman Julie Vallese.

Thus the CPSC focused exclusively on how much lead came off the surface of a lunch box when lab workers swiped them.

For the swipe tests, the results were lower, especially after the researchers changed their testing protocol. After a handful of tests, they increased the number of times they swiped each bag, again and again on the same spot, resulting in lower average results.

An in-house e-mail from the director of the CPSC's chemistry division explained that they had been retesting with the new protocol "which gave a lower average result than the prior report ... ," he wrote. "This shows ... that the overall risk is lower than our original testing would have showed, as the amount of lead dislodgeable is mostly taken out with the first wipe and goes down with subsequent wipes."

Vallese explained it this way: "The more you wipe, the less lead you actually find. With fewer wipes we got a higher detection of lead presence. We thought more wipes was closer to reflecting how you would interact with your lunch box. It was more realistic."

The test results also show that many lunch boxes were tested only on the outside, which is unlikely to be in contact with food. Vallese said this was because children handle their lunch boxes from the outside.

As a result of their tests, the CPSC issued a public statement last year reassuring consumers they had nothing to worry about: "Based on the extremely low levels of lead found in our tests, in most cases, children would have to rub their lunch box and then lick their hands more than 600 times every day, for about 15-30 days, in order for the lunch box to present a health hazard."

Vallese said the commission stands by those statements.

But the results were disconcerting to experts who reviewed them for the AP.

"They found levels that we consider very high," said Alexa Engelman, a researcher at the Oakland, Calif.-based Center for Environmental Health, which has filed a series of legal complaints about lead in lunchboxes.

"They knew this all along and they didn't take action on it. It's upsetting to me. Why are we, as a country, protecting the companies? We should be protecting the kids. I don't think in this instance they did their job."

Said Rep. Henry A. Waxman, D-Calif.: "I am concerned that the CPSC has failed to protect children from an unnecessary hazard they have known about for some time. We should protect our children by banning lead in all children's products."

Although these test results are only now being aired publicly, the CPSC did provide them to the Food and Drug Administration last summer. The FDA's reaction was completely different from the CPSC's. In July, 2006, after receiving the test results, the FDA sent a letter to lunch box manufacturers warning them that their lead levels might be dangerously high and advising them that the FDA might take action against them because the lead would be considered a food additive if it rubbed off onto kids' lunches.

"The lunch boxes containing the lead compounds may be subject to enforcement action," said the letter.

In response to the FDA warning, Wal-Mart stopped selling soft lunchboxes with vinyl liners, and offered refunds to customers who wanted to return the ones they already had.

"The safety of our customers is always a top priority for Wal-Mart," said store officials in a written statement last summer.

Other manufacturers have recently revamped their manufacturing processes to eliminate lead, or stopped making the lunch boxes altogether. Those changes have been prompted in large part by pressure from the Center for Environmental Health and several other nonprofit advocacy groups in New York and Washington State that have been testing lunch boxes and publicly airing the results for several years.

In Connecticut, where the safe threshold is 100 parts per million, Attorney General Richard Blumenthal has demanded that lunch boxes must be lead-free.

"Lead, lunch and children are a perilous mix," Blumenthal said. "The discovery of lead in children's lunch boxes is appalling. Our law is clear: Lead-laden lunch boxes are illegal."

Other states, including California, New York and Illinois, have forced specific manufacturers to pull their products from store shelves after individual boxes were found to have levels above 600 ppm.

Lead is a stabilizing agent in vinyl, but there are other chemicals that can be used instead of lead. Almost every lunch box found with lead in the vinyl lining was made in China.

But they are distributed worldwide. Other information in the documents include an e-mail from Canadian health officials, who found more than 600 parts per million of lead in seven of the 11 lunch boxes they tested.

Allen Blakey, a spokesman for the Vinyl Institute, a trade association representing the leading manufacturers of vinyl, said his organization defers to the regulatory agencies.

"The CPSC was pretty clear that they did not see a danger in these lunch boxes. The FDA had a slightly different take on it. But basically, we have not seen any indication of actual harm from the lunch boxes," he said.

Public health experts consider elevated levels of lead in blood a significant health hazard for U.S. children. Studies have repeatedly shown that childhood exposure to lead can lead to learning problems, reduced intelligence, hyperactivity and attention deficit disorder. There is no lead level that is considered safe in blood, and recent studies have shown adverse health effects even at very low levels.

"I don't think the Consumer Product Safety Commission has lived up to its role to protect kids from lead," said Dr. Bruce Lamphear, a lead poisoning specialist at the Children's Hospital Medical Center in Cincinnati, Ohio. "As a public agency, their work should be transparent. And if one is to err on the side of protecting children rather than protecting lunch box makers, then certainly you would want to lower the levels."