Air Force officials, citing improper activities at a key defense contractor, are concerned that an undetermined number of nuclear-armed, air-launched cruise missiles could fail in the sub-zero temperatures routinely encountered in war.

Congressional officials investigating the problem fear that, in a superpower conflict, cold weather could prevent some of the cruise missiles, a mainstay of the U.S. strategic arsenal, from striking vital Soviet targets after being launched from long-range bombers.

The Air Force discounts this possibility, citing an impressive success rate in cruise missile flight tests conducted over the past five years and in the readiness of its missiles to fly their wartime missions at a moment's notice.

But the service's unease was reflected by a quiet decision two months ago from the Oklahoma-based systems manager of the program to conduct low-temperature cruise missile tests costing $87,000 at Strategic Air Command bases around the nation.

The plan calls for testing every missile undergoing routine maintenance during the next year in chambers capable of simulating the extreme cold experienced during bomber flights to the Soviet Union high over the North Pole.

A more ambitious plan, favored by investigators but recently rejected by the Air Force as too awkward and costly, would have required testing a random sample of the force of 1,715 cruise missiles at low temperatures, to be followed by tests of key parts in every missile if substantial defects turned up.

The Air Force concerns stem from revelations last January that a computer-like device installed in every cruise missile to ensure stable, high-speed flight may not have been properly tested by the Northrop Corp. at a small production plant in Pomona, Calif.

The revelations have sparked investigations by the Federal Bureau of Investigation and the Justice Department's Fraud Section, which recently impounded a $1.5 million cruise missile as evidence for future legal proceedings. Northrop has dismissed four of the employes involved and decided recently to close the Pomona plant and move the remaining operations to a larger office in Norwood, Mass.

The General Accounting Office (GAO) said in a detailed report this week that a substantial portion of the cruise missiles that failed production tests did so in "very cold" temperatures. Also, two operational flight tests have been aborted because of defects that appeared in cold temperatures but disappeared when the devices in question, known as "flight data transmitters," were warmed.

Rep. Les Aspin (D-Wis.), chairman of the House Armed Services Committee that ordered the GAO report, said "the issue of cold-temperature testing is significant."

Aspin said military experts had determined that "cold is the single most stressful factor affecting the performance of the air-launched cruise missile {ALCM}. Put another way, it is the best way to find problems in the ALCM."

Eleven months after disgruntled Northrop employes disclosed the testing irregularities, Air Force officials remain unsure how many missiles contain the transmitters that Northrop did not test or did not repair after testing revealed defects.

Northrop claims a search of the testing records reveals irregularities or missing information for only 29 transmitters, which it has agreed to retest. The GAO report said however that "due to the possibility that records were falsified," the Air Logistics Center in Oklahoma City "considers all {1,715 transmitters} suspect."

Air Force spokesmen note that some of transmitters were tested again when the cruise missiles were assembled by the Boeing Aerospace Co. or delivered to Strategic Air Command bases. The missiles are routinely tested and recertified by the Air Force every three years.

Only 5 to 7 percent of the transmitters failed during these tests, the Air Force told the GAO, an ordinary rate for such parts.

But none of these additional tests was performed at low temperatures, causing one investigator to assert that "the only way the Air Force will know for sure whether the transmitters will work is when they're used {in a war} after cold bomber flights or when they break."

"The agency has a needle in a haystack problem," the investigator added. "There's really no way to find all the bad missiles without looking through the entire force."

Ten of 11 suspect transmitters tested by the Air Force at a maintenance facility this summer failed to meet official requirements, a circumstance that greatly fueled the agency's anxiety. But since then, the Air Force has decided these tests were too stringent and that the requirements should be relaxed.

This decision was based in part on a Boeing computer simulation indicating that deficiencies in the 10 transmitters were not serious enough to stop the missiles from reaching targets.

The agency's new plan was crafted to avoid undue disruptions to Strategic Air Command operations caused by taking supposedly healthy transmitters out of service for recertification. According to the GAO, the plan involves testing the entire missiles against "a less demanding set of performance standards" than specified in Northrop's production contract.

The 125 missiles involved will not be selected at random, as officials in the cruise missile systems office initially urged. There is general agreement that random tests would have allowed the Air Force to estimate with high confidence how widespread any transmitter defects were in the entire cruise missile force.

Instead, the Air Force will test missiles undergoing routine maintenance for a one-year period, beginning next spring. As a result, the tests will primarily cover transmitters produced in 1982 or 1985, relatively few of which have failed. Transmitters produced in 1983, which have a previous failure rate more than double those from 1982 and 1985, will not be included.

Aspin, citing these problems, said, "I'm not sure that we can conclude much from the upcoming Air Force retest."

The GAO stated more bluntly that the 125 tests will "only show how well the 125 {transmitters} are functioning" and "cannot be statistically projected" to other transmitters.

Gene Pickett, an Air Logistics Center spokesman, said the Air Force is attempting to get Boeing to pay for the new tests. According to the GAO report, Boeing inspectors visited Northrop's Pomona plant periodically but did not routinely observe transmitter tests or uncover testing irregularities.

Boeing officials told the GAO such "minimal checks" were commonplace and that intentional fraud was "very difficult to detect under conditions of normal. . . oversight."