Soviet and American negotiators here again are discussing how to limit their nuclear arsenals, but the best way to deal with the fears that drive the nuclear arms race won't be on the negotiating table.

That ideal solution would be to get rid of land-based, intercontinental-range missiles entirely. To be sure, this is a radical solution. But in the end it might also prove the safest and most prudent step for both sides.

The point is that the theoretical threat of a first-strike against vital military targets on each other's homeland will never really disappear until the vital targets themselves -- the underground missile silos -- disappear.

Facing up to such a proposal would be exceedingly difficult but perhaps not impossible, especially if the phaseout of these missiles was proposed to take place gradually over perhaps 10 years.

This is not a new idea. But perhaps the time has come to take a fresh look at it because the superpowers are now at a crossroads in the arms race, as the new strategic arms reduction talks illustrate, where strategic concepts should be reviewed.

The United States has had a triad of land and submarine-based missiles and bombers for about two decades now, and it has served the country well. Three different systems vastly complicate the timing any attacker would need to try to catch all three forces by surprise. It gives assurance that one or two of those forces will survive to strike back.

And, in a perverse way, the existence of missiles on land assures that Americans will die in an attack by the Soviets. In the view of strategists, this increases deterrence because Moscow will know that an American president won't accept huge casualties and thus will surely fire back if attacked.

But a world without ICBMs on American or Soviet territory would be safer than anything in the current situation. This would be a world in which neither side had large numbers of militarily useful strategic targets on the other's territory, a fact that would surely reduce the damage in the unlikely event a nuclear exchange did occur. (Today, the Soviets have about 1,400 land-based ICBMs; the United States has 1,052.)

Critics argue that reliance only on missile- firing submarines and bombers would allow Moscow to concentrate its money on antisubmarine warfare (ASW) and thus cripple the next most potent force. But the oceans are almost incomprehensibly vast and opaque places to hide in, and the United States is better at ASW anyway. Washington could sink just as much money into building even more elusive submarines and finding ways to protect them.

Communicating with submarines has also been something of a problem. But it is accomplished now and it could -- perhaps using the money saved from land-based missiles and anti-missile defenses to protect the land-based missiles -- be improved by adding more powerful land-based transmitting stations and airborne relay planes to keep in closer touch.

Critics also argue that perhaps 40 percent of our missile submarines are in port at any one time. (For the Soviets the percentage is much higher.) The answer is to build more.

The overriding point is that missile-firing submarines at sea today are all but invulnerable to attack. Thus they do not have to be fired first in a crisis and indeed are impossible for an enemy to knock out with any confidence. They offer the best way to end the fear that really grips any president and population: that there will be a hair-trigger on the nuclear button.

Along with submarines, bombers remain a useful and also non-threatening investment. They can be useful for many roles. They take hours to get where they are going and thus are not good first-strike weapons. They can carry conventional and atomic bombs. They are visible as symbols, yet they can be recalled. Most importantly, they are manned by human beings who are very resourceful.

It seems safe to predict that the idea of eliminating land-based missiles will not be considered here. Indeed, it seems out of the question. There are vast institutional pressures -- both in the Soviet and American air forces and military-civilian bureaucracies -- that undoubtedly would resist.

The thought is especially repugnant to the Soviets, who have 70 percent of their nuclear missile power on land (compared to about 30 percent for the United States.) The branch of the Soviet armed services responsible for land-based rockets has enormous bureaucratic power.

But a decade of arms control talks about more conventional solutions have not produced much success. And this proposal is potentially attractive even if the Soviets won't accept it. With a new generation of more accurate submarine-based missiles coming into service, the United States could decide unilaterally to give up its land-based missiles over a period of years, replacing them with new submarine-based missiles, thus removing these dangerously appealing targets from the maps of strategic war planners in Moscow.