The Veterans Administration has awarded medical disability benefits over the last 10 years for at least 10 veterans with leukemia or other cancers who claimed their diseases stemmed from low-level radiation exposures during U.S. nuclear weapons tests in Nevada or the South Pacific.

These claims were granted even as officials at the Pentagon and the old Atomic Energy Commission were insisting - as most continue to do - that there was no proved connection between the radiation received at the tests and subsequent cancers and other diseases.

The VA decisions are the closest the federal government has come to recognizing some responsibility for illnesses developed after troops received radiation exposure at the weapons tests.

In making the awards, the VA boards generally acknowledged only that there was "doubt" whether or not the veteran's death or illness "resulted from radiation exposure during . . . service," according to one VA report. The law provides for deciding in a veteran's favor where reasonable doubt on the cause of disabilities is found.

"We're a compassionate agency," said Sydney J. Shuman, chairman of the Board of Veterans Appeals.

Overall, however, the VA turned down a large majority of the veterans or widows who filed claims based on nuclear test exposures. While 10 were approved, at least 71 were disapproved.

Numbers are tentative because the two branches of VA handling claims - the regional offices and board of appeals - do not yet agree on how many A-test exposure cases have been filed. Records on radiation cases have only been kept since 1967.

In refusing claims, the VA decision often cited the low-level radiation doses recorded on the veteran's medical records. In one case it added that at such levels, "no injury . . . is expected."

A study of the few VA cases avilable, from these related to A-test exposures, shows a wide range of radiation doses among the claimants, both those approved and those turned down.

VA officials themselves disagree on what exposure level if any is a threshhold for disease.

In an interview last week, J.C. Peckarsky, director of the VA's Compensation and Pension Service, said, "We don't have any hard and fast rule."

He said in one of the award cases "the exposure rate was quite high . . . over five rads."

Five rads is the level considered safe by current government standards.

Appeals board chairman Shuman said, "Radiation level is not that important." He said the appeals group looks for other medical information, such as "symptoms that exist within a year of discharge."

One case, approved by Shuman's appeals board, involved a pilot whose records show he was exposed to less than one rad flying planes that collected air samples at tests in both Nevada and the South Pacific.

Six years after his discharge, the ex-pilot entered a hospital with headaches and chest wall bruises. Less than a month later he was dead of leukemia.

His widow filed for benefits and at the VA regional level, was turned down. The fact that her husband's dosage was well below the five rad level, today's safe standard, was the key factor.

At the appeals board level, however, exposure was not that important. According to the board's report, the pilot's physical decline during 1961, the year he left service, was made a focus. He "turned gray, and had to get bifocals; he aged very rapidly, seemed stooped . . ."

The board reversed the earlier decision and granted the benefits because the evidence raised "a substantial doubt as distinguished from mere speculation" that death "resulted from radiation exposure during . . . service."

Few of the turned-down claimants whose cases are available received that type of consideration.

One ex-Army man spent 13 months on Eniwetok in the Pacific at the time of nuclear tests. He was in charge of the island closest to the test area which contained the Army Signal Corps receiver station.

He observed some 20 shots from the island and after service was told his film badge indicated his total radiation dosage was 3.15 rads. In a recent affidavit he said "we often did not actually wear the badges and I cannot believe accurate readings were obtained on them anyhow."

Sixteen years after his exposure, this man was diagnosed as having cancer.

When he applied to the VA for benefits, he was turned down on grounds that his recorded radiation dosage was low and that "there was no abnormalties noted at discharge."

Another Mother for Peace, a California-based organization, has collected a series of similar stories from nuclear test veterans who have unsuccessfully sought benefits.

They include a Navy man whose records show he was exposed to 18 to 32 rads during the 1951 Eniwetok tests; a Navy diver who went into the water after the first 1946 Crossroads shot at Bikini and 23 years later died of cancer, and a Marine officer who participated in the 74-kiloton 1957 test nicknamed Hood in Nevada and died of cancer 20 years later.

In some cases, veterans were turned down because no record showing radiation exposure or even participation in tests could be found.

At recent hearings before the House Commerce subcommittee on health and environment, the Army disclosed that records of soldiers at the Nevada 1950s tests were up until now difficult to find. Often, it was said, a soldier's own personnel file did not show he went to a test on temporary duty status. The Army also said many of the individuals in some 1950s tests did not have radiation film badges and thus there never was a record of their radiation exposure.

According to Shuman, no single VA case is a precedent for another. The awards system thus will be severely challenged in the next year by an expected flood of claims from veterans exposed to radiation during the 1950s nuclear weapons tests in Nevada.

Information on more than 300 such men who later developed leukemia, other cancers or what are alleged to be radiation-related diseases has already been sent to the Atlanta-based Center for Disease Control.

CDC is studying one 1957 test, nicknamed Smoky, where troops maneuvered near ground zero after detonation of a 44-kiloton device.

To date, eight leukemia and 29 other cancer cases have been discovered in the 2,235 servicemen who took part in that test. Since only two leukemia cases should have normally occurred in a group that size, CDC officials have termed the leukemia finding "statistically significant."

For the VA, that still is not enough on Smoky. "We are waiting for someone to give us a report that there is reasonable probability," Peckarsky said last week.

His regional office has twice turned down Donald Coe, a Smoky veteran from Kentucky who now has hairy cell leukemia. Coe's appeal to Shuman's board is now pending.

Another Smoky veteran, Paul C. Cooper, is also dying from leukemia. Cooper's case first drew national attention to the 1957 test in early 1977. His regional board also turned him down. Not too long after the publicity, the VA appeals board granted him 100 percent disability - citing not his radiation but other medical signs such as a sore throat, joint aches and nausea that were noted during Cooper's Army service.

Peckarsky and Shuman bristle at any suggestion of being worred that the cost of disability payments - which more than double the average recipient's benefit check - somehow is making the VA and other agencies skeptical about finding a connection between the test exposures and later developing diseases.

"We have a whopping budget," Peckarsky said last week. "It's easier to allow benefits - and," he added, "more fun."

As long as scientists and doctors continue to disagree, however, the VA will continue to make its own decisions.