It may no longer be realistic to expect the Pentagon to build a sufficiently reliable system of national missile defense by 2005, Defense Secretary William Cohen told a Senate panel.

In testimony before the Senate Armed Services Committee on Tuesday, Cohen said two consecutive failures to shoot down a mock warhead in space with a land-based interceptor have raised doubts about the timetable.

"Perhaps it has called into question the realism of the date," Cohen said, noting that this was the view of retired Gen. Larry Welch, head of a panel that is advising the Pentagon on its national missile defense project.

Cohen repeatedly stressed, however, that both he and Welch believe the 2005 target date is "the date we ought to continue to focus on," even if it eventually slips.

Cohen is due to make a recommendation to President Clinton — probably in mid-August — on whether to proceed toward a system of 20 missile interceptors and a new high-powered X-band radar ready by 2005.

It is possible that Cohen would recommend moving the start-up date to 2006 or later. He gave no indication in his Senate testimony that he thinks this is necessary, although he pointedly left open that possibility by conceding under questioning that testing thus far has failed to show the system is technically feasible.

The 2005 date is important because, according to U.S. intelligence estimates, by then North Korea could have a missile capable of reaching the United States. The Pentagon says it would take at least five years to complete an anti-missile shield capable of defending all 50 states against a limited attack of ballistic missiles.

In remarks to reporters Wednesday in his Pentagon office, Cohen said Clinton's main choice is whether to keep the Pentagon on track toward the 2005 target date, or to foreclose that option. One potential problem with sticking to the 2005 timetable, Cohen said, is that it would put time pressure on Clinton's successor to determine shortly after taking office whether he wanted to continue on the same schedule.

"A consideration will have to be, by President Clinton, whether or not this puts any undue pressure on his successor" to decide whether to go forward with initial construction of the anti-missile system, Cohen said.

If Clinton does not give the go-ahead this fall to begin groundbreaking work next year at the X-band radar site on Shemya Island, Alaska, the entire project will be set back by at least one year, Cohen said.

In his comments Wednesday, Cohen raised the possibility that the next flight test of the missile interceptor would be delayed from October or November, as currently planned, to December. He said no decision on that had been made.

Cohen noted that it might be possible for Clinton to approve awarding the construction contracts this fall, with the stipulation that a separate decision would be made next year to actually execute the contracts. Because of the harsh weather conditions on Shemya, work can only be done during the summer.

Congress passed a law last year requiring the Pentagon to deploy a national missile defense as soon as technologically possible.

Sen. John Warner, R-Va., chairman of the Armed Services Committee, said that in his view this law amounted to a national decision to go ahead with missile defense. The only question is how soon, he said.

Cohen said, however, that determining when missile defense is technologically possible is a more complicated question involving such things as its affordability and the implications for broader U.S. national security concerns, such as arms control agreements with Russia, relations with China and European concerns.

"It's not simple. It's complicated," Cohen said. "But that's part of the work of diplomacy and persuasion, and that's what we've been seeking to do."

Carl Levin of Michigan, the ranking Democrat on the committee, told Cohen he believes Clinton should put off a decision on moving ahead with deployment, in light of the January and July flight test failures.

"The time has come to acknowledge that the 2005 deployment goal is no longer realistic, and should be adjusted," Levin said.

Sen. Robert Byrd, D-W.Va., made a similar point. "It's clear to me that we're nowhere near ready to make a decision to deploy," he said. "It would be a huge mistake" to commit now to a system that has yet to be shown technically feasible, let alone sufficiently reliable in an actual missile attack, he added.