The story of Lt. Ollie Connor haunts veteran infantrymen 35 years after the soldier's heroics gave way to disaster in a telling moment of the Korean War.

Connor, facing a herd of Soviet-made T34 tanks operated by North Koreans on a hill near Osan, grabbed a bazooka, dove into a ditch and began firing his 2.36-inch rocket launcher at the nearest steel behemoth. Positioned just 15 yards away, he landed 22 direct hits but managed to do little more than scratch the paint on the thick armor.

After dozens of American lives were lost, Connor and his 34th Infantry struggled back to their base knowing "what was going to happen when the tanks got to them," recalled retired Col. Carl Bernard, who participated in the battle. "The regiment pulled out in a panic before it fought."

Three decades later, military planners say that Connor's infantry successors would be almost as helpless against modern Soviet tanks in short- and medium-range skirmishes. The lack of an effective, one-man antiarmor weapon is considered the most important deficiency of the Army's light infantry divisions being prepared for quick response to conflicts, including those in the Third World.

The deficiency is not for want of money or attention. Hundreds of millions of dollars have been doled out in the past decade to contractors and Defense Department laboratories searching for a tank killer effective within 1,000 yards.

Five programs have been started and stopped since the Army concluded in the 1970s that its arsenal was of "marginal" use against the latest Soviet tanks. All five became victims of bureaucratic infighting, cost overruns, mismanagement or inertia, according to Pentagon officials and consultants.

The Army is getting ready to spend another $61 million to demonstrate and develop antitank technologies that it passed up in a similar exercise two years ago. The earliest the infantry can hope to have the new weapons is late 1993, officials say.

"It's a scandal," said a senior Defense Department official. "Ever since Korea, we've been told those weapons are inadequate. We've just never done that one right."

Although the Army has long-range TOW missiles capable of destroying modern Soviet armor, today's infantryman is equipped with outdated Dragon missiles as his only defense against nearby enemy tanks. Developed in the late 1960s and deployed in 1975, the 30-pound Dragon, fired from the shoulder, has little chance against newer models of Soviet tanks -- the T72 and T80 -- protected by several inches of heavy steel, military analysts said.

Dragon is "wire-guided," requiring the soldier to hold his aim on the target until the projectile explodes. But the force of firing the weapon often knocks the gunner off balance, causing him to guide the missile into the ground.

Another problem is that the Dragon's smoke and noise signal the soldier's presence as he stands exposed to enemy fire while holding the target in his sights.

Dragon has never been tested in war, but the Army contends that it would hold its own against Soviet tanks fielded in the early 1970s. But in a fight against more modern tanks, including those in the hands of Soviet allies in the Middle East, U.S. infantrymen "will shoot a lot of rounds and they will probably have very, very little effectiveness," said Brig. Gen. Wayne Knudson, the Army's director of requirements.

"If you want a highly strategically deployable force," Knudson said, "you must give them the tools to meet a reasonably sophisticated enemy who is probably going to have some pretty well-heeled armor protection on his primary combat systems.

"Unless we get something credible for our infantry, it's not going to play vitally in a future confrontation where modern-class tanks are to be attacked," he said.

Although the search for Dragon's successor began 10 years ago, it has failed to produce a new weapon despite almost continuous work by contractors, Army laboratories and the Defense Advanced Research Projects Agency (DARPA).

Officials cite the march of technology to explain the delay -- a problem, they say, that dates to the search for a longbow to defeat knights in armor. While U.S. engineers tried to design weapons capable of penetrating Soviet tanks, Soviet engineers kept finding ways to make their tanks less penetrable.

Similar advances in U.S. tank technology have blunted the best Soviet antiarmor weapons as well, the officials said.

"Each time we've done some very good [research] work on devices to kill tanks, somebody pulls off the cover and there's a thicker [Soviet] tank," said Harry Fair, DARPA's assistant director of land warfare. "Then you find out that device really doesn't make it."

But a closer look at the U.S. antiarmor program also reveals internal problems.

The effort to field a medium-range tank killer began in 1975 at DARPA with an $8 million project known as the Antitank Assault Air Defense System (ATAADS). The missile was guided by an invisible laser beam instead of a Dragon's fragile wire.

ATAADS' technology "worked like a jewel," said Maj. Gen. Ray Franklin, the Marine Corps' research director who worked at DARPA at the time. Ten missiles were successfully demonstrated.

But the Army, balking at the high costs of ATAADS, decided that "there really wasn't an urgency to replace Dragon," Franklin said.

Other sources said they think the Army shunned the technology because it was not invented by the Army's Missile Command, which develops and procures missiles for the service. DARPA, a Pentagon-wide laboratory, works for all the services. "The Army took the technology and it just disappeared," Franklin said.

By 1976, the Army began focusing on a shorter-range weapon called Viper, which became its costliest failure among tank killers. It was supposed to be a lightweight, shoulder-fired rocket that infantrymen could use to stop a tank within 300 yards. General Dynamics Corp., the contractor, offered to sell Vipers for $78 apiece.

Seven years later, Viper's cost had soared to nearly $1,000 each. Its cloth sling alone cost as much as the original estimate for the entire weapon, according to Army documents.

Two years behind schedule and its shell determined to be too small to kill tanks, Viper was canceled in October 1983, after the Pentagon had spent $250 million.

Today's soldier still carries the Light Antitank Weapon (LAW), but it was considered useless against tanks shortly after its introduction in the 1960s.

The Army revived its efforts for a medium-range weapon in 1980 under the name Infantry Man-portable Antiarmor Assault Weapon System (IMAAWS). The program focused on two technologies: the laser-beam concept adopted by the Missile Command and a recoilless rifle promoted by the Army's Picatinny Arsenal.

DARPA, meanwhile, was working independently on a concept, called Tank Breaker, that sought to free the gunner from the guidance process. Instead, an infrared "seeker" would direct the missile and lock it onto the target.

With Tank Breaker in the experimental stage and the need for a new weapon considered urgent, the Army awarded $35 million to two contractors to demonstrate the laser-beam and recoilless technologies. Contracts were issued Sept. 12, 1980, with the understanding that they would last two years.

Six weeks later, the Army canceled the contracts in what retired Lt. Col. Rick Briggs, then the systems coordinator for replacement of Dragon, said amounted to bureaucratic euthanasia.

"We had the Missile Command bureaucracy, the Picatinny bureaucracy and the DARPA bureaucracy all thinking they had the solution," Briggs said. "After we awarded the contracts, DARPA officials convinced a number of people in OSD [Office of the Secretary of Defense] and on the Hill that the Army was throwing away its money and that the only way to go was with the Tank Breaker."

Amid intense lobbying, Briggs said, the Army's vice chief of staff "had to decide what we were going to do. He said, 'Let's cancel the contracts and go back to studying it.' Nobody could sort out the mess." The money was simply parceled out to the three laboratories to further investigate their technologies, he added.

By 1982, the Army had a new name for its medium-range antiarmor program, Rattler, and a new technology to consider formally, Tank Breaker, whose "fire and forget" system was showing great promise. The DARPA missile also had the advantage of arcing toward its target, hitting the tank at a more vulnerable spot than with the straight-line path of Dragon or the laser-beam system.

The Missile Command, meanwhile, was still pushing its laser-beam weapon in what became another destructive bureaucratic contest for the Army's research dollars and interest, according to officials.

Rattler funds were given to both agencies to demonstrate their technology. As the research efforts wound down in 1983, however, the infighting surfaced. DARPA presented the Army with a technical study concluding that the laser-beam weapon was not sufficiently lethal against Soviet armor. The Missile Command countered with charges that Tank Breaker would be unaffordable.

The Army, faced with these new findings, decided to defer the issue again. It canceled Rattler in January 1983 and ordered new cost analyses and technical assessments.

An Army analyst reviewing the antiarmor program in June 1983 wrote that with Viper too costly and a medium-range weapon "literally being studied to death . . . it appeared that the aging Dragon and LAW would remain in the Army inventory well into the 1980s."

Critics contend that at least part of the delay is caused by the Army's institutional bias against technologies originating elsewhere.

For example, Tank Breaker, which has received more than $30 million in research funds, was successfully demonstrated in 1983. "The Army sees Tank Breaker as an outsider and has a hard time accepting the idea that it wasn't invented here," said a Pentagon consultant.

Knudson asserts that bureaucratic red tape rather than bureaucratic jealousy has delayed the search for a new tank killer. "All I want is something that works," he said.

After two years of cost and technical analyses, the Army initiated yet another program called Advanced Antitank Weapon System-Medium (AAWS-M) with plans for a $61 million demonstration and "shoot-off" of competing antiarmor technologies, including the two systems deferred in 1983 -- Tank Breaker and laser-beam guidance.

Congress, in appropriating the funds, noted that "Army management of several antiarmor programs in the past has been generally deficient, both in terms of adequate exploitation of newer technology and of obtaining cost-effective and affordable weapons."

Knudson said he thinks that the problems are behind the Army: "We would like to believe we would be issuing that stuff to the first units in late 1993 or early 1994."

Carl Bernard, the Korean War infantry veteran who is now a defense consultant in Alexandria, and other critics are less optimistic. To them, the past decade of false starts reflects a historic failure by the Army to equip U.S. infantrymen adequately against tanks. The problem, they said, dates to World War II, when U.S. troops had to use captured German weapons against German tanks because their 2.36-inch bazooka was ineffective.

The same 2.36-inch rocket launcher was issued to Lt. Connor and his 34th Infantry in Korea, replaced in the 1960s by the ineffective LAW and in the 1970s by the troubled Dragon.

"We're in the same situation today," said Bernard. "Nobody gives a good goddam about the infantry.