An Air Force investigation of six of the nation's largest defense contractors in 1983 and 1984 found evidence of pervasive inefficiency and sloppiness in the production of aircraft, missiles and other weapons, according to documents obtained by The Washington Post.

The investigation of about five weapons systems produced by each of the contractors found that the companies frequently took two or three times longer to build weapons than called for by the contractors' own work standards. Much of the extra effort went into components that failed inspection and had to be reworked or discarded.

The studies suggest that the Defense Department could buy the same number of weapons for substantially less money if contractors operated closer to their own efficiency targets. The documents also support a 1983 assertion by Paul Thayer, then deputy defense secretary, that shoddy work adds 10 to 30 percent to the cost of U.S. arms.

Richard D. DeLauer, former undersecretary of defense for research and engineering, last September warned a group of contractors that "the No. 1 weakness" of the Defense Department "is the poor workmanship" common to a great cross-section of weapons rolling off assembly lines.

"What we have is just crappy, shoddy workmanship," DeLauer said, according to Defense Week.

Industry spokesmen and some Pentagon officials, however, say the Air Force studies paint an incomplete picture of contractor efficiency and competence. They argue that the documents fail to account for recent changes intended to sharpen productivity or for inefficiencies caused by government requirements, such as frequent changes in weapons design.

The Defense Department's alarm over the productivity and quality of its arms makers came to a head more than two years ago after an investigation of the Hughes Aircraft Co. plant in Tucson uncovered "undisciplined work habits and lax management," according to congressional testimony by Major Gen. Bernard L. Weiss, chief of Air Force contracting policy.

The Air Force found that Hughes took 17 times longer than the company said it should to build the Maverick missile. Air Force auditors also concluded that Hughes "perpetuates inefficiency by hiring four to five times" as many workers as it needed, documents show.

After the military temporarily stopped accepting its missiles, Hughes devised a "get-well" plan that the Air Force says should correct the shortcomings.

But the problems at Hughes prompted Air Force Secretary Verne Orr in May 1983 to order studies of five other major firms: the Boeing Co., General Dynamics Corp., Lockheed Corp., the Pratt & Whitney division of United Technologies Corp. and Raytheon Corp. The Air Force found serious inefficiencies in at least some weapons systems being produced at all five firms, according to the documents.

Air Force officials declined to make the studies available, saying they could be "misinterpreted" or prove embarrassing to the contractors. The documents, briefing papers of 25 to 35 pages each, were obtained independently.

They show that at Raytheon, for example, workers took almost four times longer than the company estimated they should to build Aegis radar components for Navy ships. For every 100 hours the company should have needed to build a radar, almost 35 were spent building parts that had to be scrapped or repaired, Air Force documents show.

A Raytheon spokesman had no comment last week on the study, except to note that the Aegis is "a complicated shipboard system."

Similarly, on a contract to build J52 engines for Navy A6 planes, Pratt & Whitney worked three hours for every hour that its engineers said it should during 1983, a worse record than it compiled five years earlier on the same engine. The company spent almost half as much time on scrap and rework as it should have spent building the engines.

Tom Drohan, Pratt & Whitney's vice president for public relations, noted that the Air Force study also had praised the company for its "commitment to productivity improvement," including plans for greater capital investment. He said the actual time spent on reworking parts of the J52 is a tiny fraction, about 2 percent, of the total effort on the engine.

The Air Force study showed, however, that scrap, rework and repair accounted for 15.5 percent of the hours Pratt & Whitney devoted to the J52 engine and for 46.1 percent of the hours the company would have spent if it had worked at target efficiency.

General Dynamics took two hours for every hour its engineers said it should on its Tomahawk missile and KC10 refueling aircraft, and almost four times longer than its efficiency target on Atlas rocket contracts, according to the Air Force study completed in early 1984.

"Since 1983, considerable improvement in performance, scrap and rework has been made across the board on the Tomahawk, KC10 and Atlas programs," General Dynamics spokesman Alvin Spivak said.

"Our overall conclusion was that, on some of these programs, certainly the contractors were not as efficient as we'd like them to be," said one Air Force official, who asked not to be identified. "So we've adjusted our policy to negotiate a very tight budget at the outset of a contract to force the contractor to stay within that budget."

All five companies, including Raytheon and Pratt & Whitney, performed much better on some systems, and Lockheed Corp.'s Georgia plant performed with relative efficiency on almost all contracts. None of them reached their efficiency targets, however.

As a result of the study's findings, Gen. Lawrence A. Skantze, commander of Air Force Systems Command, last September ordered the Air Force to negotiate contracts based on what a weapon should cost, not on what history shows it is likely to cost.

"We're not going to pay for previous inefficiency anymore," one of Skantze's aides said. The official said it is too soon to say whether the new policy is having an effect.

The official also said that companies tend to perform better on mature programs, weapons without complex electronics and programs in which the Pentagon is buying sufficient numbers of items. Boeing, for example, performed poorly on the Roland missile, which was canceled early in its life, but came close to its target efficiency on the air-launched cruise missile.

Boeing spokesman Bill Rice said that under Pentagon procurement rules, most firms establish a "standard" hour, which is based on calculations by the company's industrial engineers of how much work an efficient plant should accomplish. "The potential to attaining the standard only comes after working on the program for some time," Rice said.

For example, he said, Boeing took 5,500 hours to build the first air-launched cruise missile, but only 1,000 hours to build missiles that came off the line a year later.