GEN. BERNARD W. Rogers, soon to depart as NATO's military commander, says that without nuclear weapons the alliance could resist a Soviet attack only for "days, not weeks." The more relevant question is how long could Western Europe survive if NATO used nuclear weapons to resist a Soviet attack?

Rogers makes the same mistake that many military strategists and governments make in talking about nuclear arms: They talk about them as if they were traditional weapons that can be used to fight and win wars.

However, nuclear devices are simply not weapons, for the use of even a very small number of them would destroy the battlefield -- that is, Western Europe and perhaps most of the world. No one knows how to fight a war using anything like the thousands of nuclear devices that both the United States and the Soviet Union have in Europe.

NATO planners have been trying for 30 years to come up with a sensible plan for using nuclear weapons in Europe and have been unable to do so. The current plan calls for firing one or a few weapons -- the shot across the bow -- in the hope that the Soviets will be sufficiently scared to sue for peace. However, many analysts believe that it is much more likely that the NATO populations will demand an end to a war on any terms if a nuclear device goes off.

Now the Soviets are proposing to eliminate two categories of nuclear weapons: medium-range and short-range missiles. The agreement in principle of the NATO countries is what has Rogers and other critics disturbed.

What NATO now relies on to deter a Soviet conventional invasion of Western Europe is a doomsday machine linked to a roulette wheel. In other words, the alliance says to the Soviets: There are so many nuclear weapons lying about in Europe that if you start a war one of them may go off. And, if one weapon explodes, many more are likely to be used. The result would be the destruction of the world.

People like Rogers, who understands how irrational and dangerous the present system is, become nervous about any proposed change in nuclear deployment in Europe whether by agreement or unilateral action that would give the Soviets less reason to fear the doomsday machine.

This unease is often expressed as a fear of "de-coupling." Rogers and others argue that unless the United States has one or another nuclear delivery systems in place in Europe and maintains the right ratio of NATO and Soviet weapons, the Russians are likely to conclude that the American president would not use nuclear weapons to resist a Soviet conventional attack in Europe. However, they don't make clear why an American president is more likely to use medium-range missiles based in Europe -- rather than weapons offshore or in the United States -- to attack Soviet forces.

In fact, Washington is likely to believe that any large-scale nuclear attack on Soviet forces would lead to a massive Soviet response aimed at all American forces in Europe as well as at the rest of the American nuclear capability, including forces in the United States. Thus, if an American president ever decides to initiate a nuclear attack, he is likely to order an all-out strike designed to destroy the Soviet leadership and as much of its nuclear capacity as possible, not a partial strike aimed at some Soviet forces in Europe.

A more fundamental concern is that the Soviets might conclude that the president would not initiate the use of nuclear weapons regardless of what delivery systems exist. Because nuclear devices are not weapons that can be used to fight wars, and because any use of these devices runs the grave risk of destroying the world, there is no way to make the threat to initiate their use credible.

Despite the serious risk that the current deployments and war plans could lead to a nuclear war that neither side intends, those who oppose the agreement and support the status quo argue against change for two fundamental reasons. First they say that the system ain't broke and should not be fixed. Second, they argue that there is no real alternative. Neither argument is correct.

It is, of course, true that Europe has escaped war for the past 30 years. But we are courting danger every day. Consider just the following: The military has been told that it will be given permission to use nuclear weapons whenever "necessary." In a crisis brought on by a Soviet ultimatum or a Soviet military move following unrest in East Europe, NATO might suddenly find that it was not willing to resist a Soviet conventional military move because of the belief that within a few day it would have to initiate the use of nuclear weapons or be overrun. Ironically, the pressure to appease the Soviets in order to avoid war could lead to the neutralization of Western Europe that those who resist change most fear.

Alternatively, in a crisis NATO leaders might come to realize how vulnerable their nuclear weapons, stored in few well-identified storage sites, were to a Soviet conventional attack, and give in to pressure to disperse them. Once the weapons were deployed, the danger of unauthorized use or approved use to prevent their being overrun would increase.

If the current situation poses too high a risk of an unintended nuclear war or of the NATO alliance backing down in a crisis, we must find an alternative.

There is one. If nuclear weapons had never been invented, we would have found a way to protect our security interests from the Soviet Union. That same option is available now. We cannot get rid of nuclear devices nor entirely remove the risk that a war in Europe would go nuclear. What we can do is to make our own war plans and deploy our forces on the assumption that it would never be in the interest of NATO to initiate the use of nuclear devices.

We can deploy the nuclear weapons in a separate command so that devices are available to retaliate for Soviet use or in the unlikely event that NATO seeks to initiate their use. The regular military forces would be trained and equipped only for sustained conventional war, rather than having forces that seek to be dual-capable. This change in doctrine would enhance NATO's conventional capability with no increase in defense spending. Deterrence would thereby be more credible.

At the same time we would recognize that what effectively deters the Kremlin is the credible threat that a war in Europe means a large war with the United States that we will fight to win. As long as no events in Western Europe threaten vital Soviet interests, the Russians are unlikely to attack in this situation.

Having recognized that nuclear devices are not weapons and are no substitute for an adequate conventional capability, we can have a serious debate about whether current NATO forces are adequate to deter a Soviet attack. My own sense is that with modest improvements they are surely sufficient. Those who think not should be making the case for the needed improvements and not misleading us into thinking that we can rely instead on nuclear weapons if only we reject the current Soviet proposal.

It is time for realism in the alliance. That must begin by recognizing that nuclear devices are not weapons. Agreements to cut them back cannot threaten our security. What can is the false belief that we can rely on them to deter or defeat conventional threats.

Morton H. Halperin, a deputy assistant secretary of defense in the Johnson administration, is the author of the recently-published "Nuclear Fallacy."