ASHINGTON, May 8 — Some soldiers, sailors and aviators who developed cancer from exposure to radiation from 1945 to 1962 were denied compensation because the Pentagon grossly underestimated their doses, a panel of independent scientists said today.

For a majority of veterans who took part in cold war nuclear tests or were in Japan near Hiroshima and Nagasaki, the underestimation does not matter because "ionizing radiation is not a potent cause of cancer," said the panel, which was convened by the National Academy of Sciences at the request of Congress.

Congress has classified 21 kinds of cancer as "presumptively" caused by radiation exposure. About 4,000 veterans with other kinds of cancer or other diseases applied for compensation, and all but around 50 were turned down, the study found.

The study's authors said they could not estimate how many of the others should have been compensated. "Let me emphasize how difficult it was to even sort out this number of 50," said John E. Till, committee chairman and president of the Risk Assessment Group of Neeses, S.C. "It is impossible for us to say how many claims might be successful should these claims be recalculated." But it was appropriate to reject most of the 4,000, the report said.

It was unclear whether the doses of unsuccessful claimants would be recalculated. Mr. Till said this was outside the committee's assignment.

Lt. David Guy of the Navy, a spokesman for the Defense Threat Reduction Agency, which made the calculations in the first place, said that the agency was in general agreement with the report but that it would use it to reform its procedures, not to revisit past work.

Mr. Till's committee, the National Academy of Sciences Board on Radiation Effects Research, stated that in some of the 99 cases it reviewed in depth, the calculations were illegible or unexplained. In other cases, dose analysts ignored the possibility that a blast at the Nevada Test Site would kick up fallout deposited in previous tests, the panel said. And information from the veterans about their activities at the test scenes was often ignored, the reviewers said.

In one case, a major who said he was present at 21 detonations was credited with having been at only 11.

Congress intended the dose reconstruction process, which, by definition, is an estimate, to give the benefit of the doubt to the veterans, and told the Pentagon to calculate the maximum possible exposure for each veteran, and use that as the working figure.

Veterans were to be compensated if the probability was 50 percent or more that the exposure was the cause of their disease. But the reviewers said that in many cases the Pentagon's estimate was 10 times too small.

The question of "atomic veterans" has persisted for more than 20 years, but as the debate has continued, the number of veterans has dwindled. Of those covered in the study released today, the oldest were prisoners near Hiroshima and Nagasaki in 1945 or stationed there after the war. The youngest were those exposed in the last days of atmospheric nuclear testing, in 1962.

William A. Harper, commander of the National Association of Atomic Veterans, said in an interview that "if guys could get some kind of compensation out of it while they are still living, that would be nice." His group has fewer than 5,000 members, down from 10,000 at its peak, he said.

Mr. Harper, 77, was a Navy petty officer in the South Pacific during two nuclear blasts in July 1946. He developed polio a few years later, and said it was caused by radiation's effect on his immune system. He was turned down for compensation.

Mr. Harper said that the Pentagon had applied a single dose estimate to everyone on a ship, even though sailors had different jobs that resulted in differing exposures.

An independent radiation expert, Arjun Makhijani, who in 1983 published a critique of dose estimates from the July 1946 tests, said the government should simply provide compensation and medical care to the surviving veterans.

But a member of the committee, Clarice Weinberg, chief of the biostatistics branch of the National Institute of Environmental Health Sciences, said that the recognition that the dose estimates were poor was not the same as saying that they were high enough to cause cancer. "Even at those levels of exposure, radiation is not that potent as a carcinogen," Ms. Weinberg said.

Studies of the Japanese at Hiroshima and Nagasaki exposed to far higher levels found that only about 5 percent of the cancers they suffered were a result of radiation, she said.

For the veterans, she said, "For many of these doses, you could multiply by 10 and even 100, and not come up to a level that would warrant the claim being awarded."