The M1 tank is still beset by breakdown and by repair and other serious problems that could render it ineffective in combat, according to internal Army test reports released today by a civilian Pentagon watchdog group.

The Army has known about the troubling test results since last fall but apparently delayed releasing the data outside the Pentagon until after Congress voted to authorize $1.3 billion for procurement of 855 of the tanks and authorized another $380.9 million for advance procurement of 900 more M1s, said Dina Rasor, who released the Army documents.

Rasor is director of the independent Project on Military Procurement.

The Army also manipulated data to make the tank look much more successful than it is, Rasor said. As recently as last March, the Pentagon assured Congress that the tank, each of which costs $2.7 million, "was the best tank in the world" and quoted figures that its own tests did not substantiate, the documents show.

The test results show that the durability of the Chrsyler-made M1 actually has worsened since it was tested in early 1980. Repairs also continued to prove difficult even under peacetime conditions.

The documents also show that some M1 crews have switched off one of the tank's sophisticated gadgets, the muzzle reference system, during training exercises because they believe it caused the tank's main gun to fire inaccurately.

Rasor said the documents also show that hydraulic fluid used in the tanks, which M1 critics claimed is highly flammable, has been linked to at least two tank fires and that intense heat produced by the tank's turbine engines makes it difficult for the M1 to hide from an enemy equipped with forward infrared thermal imagers such as those being installed by the Soviets on their newest weapons systems.

According to the documents, M1 hatches leak in the rain and the .50-caliber machine gun is "susceptible to accidental firing" -- a complaint made during earlier tests in 1980 but apparently not resolved.

Rasor said the most troubling aspect of the tests, supervised by the Army's Test and Evaluation Agency, was the tank's durability rate. The Army has said that 50 percent of the M1s must be able to go 4,000 miles without an engine, transmission or final drive replacement or overhaul.

In March, Army officials told Congress that 37 percent of the tanks tested could achieve that goal. But, Rasor said, test results show that the 37 percent figure is a combination of field testing and laboratory testing.

By combining the two test scores, the Army enhanced the tank's durability rating, she said. In the field, only 15 percent of the tanks could go 4,000 miles without breaking down, the test showed.

The documents also revealed results of a 96-hour field exercise at Fort Hood, Tex., that the Army said would "match or exceed the most stretchful [sic] and arduous mission" faced by the M1 during combat.

The exercise began with 39 tanks. Five days later, only 21 had completed the 178-mile course without breaking down, which means that nearly half of the new tanks became disabled during the exercise without facing enemy fire, Rasor said.

Last December, the General Accounting Office urged Congress to delay large-scale M1 production until the tank's power train could be made more durable. GAO said the power train failed after an average of 36 miles during testing at Fort Hood.

At congressional hearings in July, Maj. Gen. Richard Lawrence, commanding general at Fort Hood, blasted the GAO report, saying the agency did not base its report on any test data from Fort Hood. The Army's internal tests released by Rasor are identical to the GAO figures.