For some years now, the Pentagon has insisted that the best way to defend against enemy ballistic missiles would be to fire volleys of ground-based interceptors.

Space-based interceptors went out of fashion -- and the realm of affordability -- with former president Ronald Reagan. And laser devices, popular with some missile defense enthusiasts, have a long way to go before they are ready for prime time.

But the ground-based interceptor approach wasn't working very well either -- until lately. After years of suffering many more misses than hits, the Pentagon has scored a string of successes in recent months shooting at mock enemy missiles using a new generation of prototype interceptors.

The latest triumph came Saturday night when the most advanced of these model devices -- the one that would be used to defend all 50 U.S. states against missile attack -- flew its first intercept attempt and pulverized a dummy warhead about 140 miles above the central Pacific Ocean.

The test involved higher altitudes and missile speeds at least three times faster than earlier intercepts this year by two Army prototypes -- the Patriot 3 and the Theater High-Altitude Area Defense, or THAAD -- meant for shorter-range battlefield systems. Together, these trial runs have provided critical validation of the "hit-to-kill" concept pitting one speeding missile against another.

But can these interceptors succeed every time?

While ecstatic over the successful Pacific test, defense officials cautioned yesterday that it marked only a first step, with many more to go before the United States could be confident of having a reliable protective umbrella. Critics, having decried the effort to build a national shield as unworkable, risky, too costly and dangerously disruptive of relations with Russia and China, appeared not about to be silenced.

One leading skeptic -- John Pike of the Federation of American Scientists -- granted that Saturday's hit, even under the test's carefully controlled and limited conditions, was no small achievement. But he remained doubtful of the system's reliability under more stressful combat conditions.

"What they've done is the equivalent of shooting a hole-in-one," Pike said. "What they have to be able to do is shoot a hole-in-one every time. Missile defense must work perfectly if it's going to work at all. They can't afford to miss."

John Steinbruner, a senior fellow at the Brookings Institution, worried that the U.S. effort would so antagonize Russia and China that any hope of further nuclear arms cuts would be shattered.

"Everyone says we'll fix it with them, but we won't," he predicted.

To deploy the national shield as planned, the United States must win Russian agreement to amend the 1972 Anti-Ballistic Missile Treaty, or it must withdraw from the accord. Hoping to improve prospects for a deal, the Clinton administration has decided to limit the initial deployment to 100 interceptors at a single site in Alaska. Plans to establish a second site and expand the number of interceptors to more than 200 would be left for a second phase of negotiations at some later unspecified time.

This two-phased approach hasn't done much yet to budge the Russians. At the same time, it has angered some congressional Republicans, who want either a more thorough overhaul of the ABM Treaty, which they regard as a Cold War relic, or its dissolution.

Sen. Kent Conrad (D-N.D.), whose state has been vying with Alaska to house the interceptors, also is upset with the administration's plan to start with just one site. He said two sites would allow more follow-up shots at incoming warheads if the first volleys miss.

"With only a single site in Alaska, you particularly diminish the capability of defending the East Coast," Conrad said.

Pentagon officials say the Alaska site would provide adequate coverage for all U.S. states. But the administration clearly is trying for some middle ground between a two-site scheme, which the Pentagon would have preferred, and no scheme at all, which the Russians maintain is still their position.

Republicans have faulted the Clinton administration for moving too slowly to develop a national missile defense system, and several GOP presidential candidates have made it a major plank in their platforms. But the program already is moving about as fast as the contractors involved say it can, given the technological hurdles.

Last January, the Pentagon pushed back the earliest estimated deployment date, from 2003 to 2005. And President Clinton has yet to make a final decision to build the system. He is due to review the program next summer, after a total of three scheduled intercept tests. Pentagon authorities had hoped for four tests by then, but a three-month delay this summer has made that unlikely, officials said.

With so much riding on Saturday's test, conditions were tightly circumscribed to increase the chances of success.

The target was simplied to include fewer decoys than existed in two earlier fly-by tests. And both the target (a modified Minuteman launched from Vandenberg Air Force Base in California) and the intercepting "exoatmospheric kill vehicle" (sent aloft atop a booster rocket from the Kwajelein Missile Range in the Marshall Islands) were maneuvered onto a collision course with the help of global positioning satellites in the test's early minutes.

But the kill vehicle, once released from its booster, received neither external assistance to distinguish the dummy enemy warhead from a decoy balloon nor any help to close in on the warhead, officials said. Neither object carried an explosive. They were destroyed by the energy of the collision, which occurred at a combined speed of about 15,000 mph -- fast enough, one official said, to go from Washington to New York in a minute.