The United States builds almost twice as many nuclear warheads of each design as it plans to deploy according to nuclear weapons officials.

The excess warheads are used up in the extensive stockpiling sampling rquired annually to catch the many non-nuclear mechanical and material aging problems that develop over the 20-year life of each nuclear system.

"Reliability of the nuclear stockpile rests on the maintenance program," a top official at the government's nuclear weapons laboratory at Livermore, Calif., said recently.

In the annual stockpile tests, the nuclear component is removed from the randomly selected warheads of each weapon and the remaining parts are tested under simulated operational conditions.

It was these tests that in 1965 first turned up the locked safety device on the then five-year-old Polaris Al sub-launched nuclear warhead. Additional tests in 1966 proved to scientists that because materials in the safety device deteriorated three-quarters of the Polaris warheads then at sea would not have worked.

It was learned last week that the 1966 problem was not the only one suffered by the Polaris.

In the early 1970s, according to informed sources, an aging organic material in the warhead created a vapor that "affected a critical part." Had it not been discovered and remedied, the critical part would have eroded and prevented the warhead from operating, sources said.

Department of Energy officials refused to comment on the 1970s Polaris incident, saying "it was too close to the present time."

Other officials, however, confirmed that "over the long term it could have led to no confidence in the Polaris missile."

Problems develop in nuclear warheads "every once in a while," a weapons scientist said recently, "because of the incompatibility of materials."

Each weapon contains unstable chemical high explosive material that is needed to trigger the nuclear explosion. In addition, the warhead must have nuclear material, which gives off radioactivity. Other materials used in the device to seal parts and hold them together can in turn be affected by either aging or the radioactivity.

While the non-nuclear parts of each warhead design are tested each year, the nuclear component is never tested once the device is certified at production time.

One understandable weakness in the entire nuclear program, according to interviews with present and former officials, is that the United States has never tested one of its current land-or sea-based intercontinental missiles with a live nuclear warhead.

As far as is known, these officials qucikly add, the Soviets have never made such a test either.

"What gives less confidence to our strategic force," a former Pentagon official said recently, "is that they are untried weapons of immense complexity."

Officials believe that the Soviets probably run into the same aging problems with their weapons. A few scientists say, however, that because the Soviet warheads are simpler and larger they don't have quite as many difficulties.

Accounts of past problems with the U.S. nuclear weapons stockpile -- such as the temporary failure of the Polaris A-1 -- have surfaced recently as part of the debate over a comprehensive nuclear test ban.

At present, the United States, Soviet Union and Great Britain have agreed not to test underground devices of more than 150 kilotons (the equivalent of 150,000 tons of TNT). Negotiations are under way for a new agreement that would drop that level to zero.

Opponents of such a total ban have argued that some underground testing is necessary to preserve the reliability of the U.S. nuclear arsenal.

Supporters have argued that weapons today in the U.S. nuclear stockpile are not subjected to nuclear testing for reliability -- only to severe testing of the non-nuclear parts of each system.