THE CONFEREES on the defense bill have a difficult and telling decision to make on the future of AMRAAM, the advanced medium- range air-to-air missile that the Air Force and to a lesser extent the Navy have been banking on as a weapon of the future. The missile, which a pilot would fire when his target was still a radar blip, is in trouble. Some people think it is conceptually flawed. Technical problems continue to crop up as well; there have been long delays in designing and preparing to build it. The projected cost has more than doubled in the past four years, to $10.8 billion for the program, $400,000 per missile. In January, Defense Secretary Caspar Weinberger took the unusual step of putting the new weapon on probation, saying no missiles would be bought until the Air Force satisfied him it had costs under control; a review is scheduled later this year.

The House responded to these uncertainties by voting to kill the missile. The Senate, brushing aside the problems, appropriated funds instead to begin production. The conferees must choose -- but more is involved than just another weapon.

The new missile represents a major commitment on the part of the Air Force, a leap of faith as to the nature of future warfare. The service has come to the view that a lot of the serious fighting will be done at much greater distances than before, even over the horizon. It is gearing up accordingly.

Its critics have a more conventional -- they would say more practical -- view of the future. The new missile has this major problem: a pilot currently has no foolproof way of telling whether a blip on his radar screen is friend or foe. Proponents are sure that a way will present itself; they say that, if one did not, in an all-out war pilots would simply be told to assume that planes in certain sectors were enemy, and to shoot without inquiring. Not a comforting answer. Nor was it reassuring to learn from a leaked document last week that the project officer does not think the manufacturer, Hughes Aircraft Co., is able yet to produce the weapon.

Congress should not appropriate production funds for a weapon of this importance under circumstances as shaky as these. A production vote would further undermine public confidence in both the military procurement system and Congress' own judgment. The conferees should either kill the project or keep it barely alive and give the Air Force a last small chance to justify it. No one can have much confidence in the weapon so far.