After balking for almost six months, the Army has agreed to conduct realistic vulnerability tests next year using live Soviet ammunition against combat-equipped, new M2 Bradley armored troop carriers, according to James P. Wade Jr., principal undersecretary of defense for research and development.

The agreement to conduct the tests came one day after the release of Defense Department documents by Rep. Denny Smith (R-Ore.), including one that blistered the Army's internal testing unit for refusing to conduct the tests with the realism originally proposed.

Earlier, in a Sept. 27 letter to Defense Secretary Caspar W. Weinberger, Smith and three other legislators asked for "a thorough investigation" of a reported attempt to remove from his job the person who had vigorously pressed for the realistic tests, Air Force Col. James G. Burton.

The legislators also argued that the incident reinforced the need for the congressionally mandated Defense Department office of operational test and evaluation. President Reagan has yet to appoint a director for that office although the legislation passed a year ago.

Burton had been nominated for the job by the sponsors of the legislation. Weinberger's initial nominee, Federal Aviation Administration associate director Donald R. Segner, was proposed last May but the White House never approved the nomination, according to the released Pentagon documents.

"No current front line U.S. system, armor or aircraft except the A10 has ever been tested through a formal test program in a combat configuration for vulnerabilities to live Soviet weapons," Wade, who is Burton's boss, wrote in an Aug. 24 memo.

"Full-up testing is urgently needed," Wade said, "to determine and correct where possible the vulnerabilities . . . . "

Plans for testing against aircraft were satisfactorily worked out with the Air Force, Wade wrote, but the "armor community," he added, "has been unwilling to prepare a plan consistent with the objectives" of the program.

An Army spokesman said recently the disagreement has been over when, not if, such tests should take place. "We always planned to do these tests in 1986," he said, "knowing that large-caliber, shaped-charge shells would destroy the Bradley."

In June, according to one of the documents released by Smith, Burton sent Wade a memo that sharply criticized the test program set up for the million-dollar Bradley vehicle. Up to that point, the tests had been run by Ballistic Research Laboratory (BRL), the Army's internal testing unit.

BRL, Burton wrote, "selectively reduced the severity" of the testing. The few shots that went off were aimed at spots on the Bradley vehicle that would cause neither an explosion nor a fire.

When BRL did one of the shots Burton demanded, "the results were devastating," he wrote. "Pressures and temperatures were double" what had occurred on the controlled shots, and "all hatches blew off, including the rear ramp."

"Without completing the tests for the real vulnerability of combat-loaded vehicles," Burton wrote, "insights on how to improve the survivability of these vehicles will be denied."

Wade said that the Army had agreed to conduct "at least 10 tests in 1985" using fully-loaded Bradley vehicles. And although the tests would not be under his direct control, the Army would permit "onsight observers" to see that they were correctly done.