THE INF AGREEMENT signed by President Reagan and General Secretary Gorbachev, though helpful to U.S. security, is only a sideshow. START remains the center ring.

Moving beyond this week's summit discussions to a full-fledged strategic arms reduction treaty requires recognition of three realities:

The current outlines of the START agreement are about as good as can be expected and deserve strong support from arms reduction adherents in both parties;

The U.S. needs to restructure its retaliatory forces so that they will still provide stable deterrence of a Soviet attack after START force reductions;

Papering over disagreements about strategic defenses is a prescription for real trouble down the road.

The opponents of arms control -- most of them Republicans -- will, while grumbling about INF, reserve their main fire for a START agreement and the limitations on strategic defense (implicit, explicit, or through reaffirmation of the ABM Treaty) that will go with it. All the more reason for supporters of arms control and reduction -- especially Democrats -- to give due credit to the Reagan administration. Ronald Reagan's conservative and anti-Soviet credentials should make it easier for his successor, of whichever party, to include arms control and U.S. military force structure in a single national security policy.

Despite the absence of breakthroughs in last week's summit, the START agreement that is shaping up is, on balance, a good one for U.S. security. A limit of 6000 ballistic and cruise missile warheads is lower than most of us thought achievable in the early 1980s. Reducing by half the number of SS18s -- currently the biggest, most accurate ICBM in the Soviet arsenal and hence the ones most suitable for a preemptive strike -- is a significant achievement. The sub-limit of 4900 ballistic missile warheads will help stability, though a further limit on ICBM warheads would be even better. Counting rules for missiles or other warheads carried over from SALT II or Reykjavik are a reasonable compromise from the U.S. point of view. The proposed limit of 1,600 launchers would allow enough leeway for deployment by both sides of single-warhead, mobile ICBMs.

As in the case of INF, these verification procedures for START will be crucial in obtaining political acceptance as well as in assuring detection of violations that would sig-nificantly affect the strategic balance. Some cooperative procedures will probably be necessary, especially for mobile missiles; here the INF verification procedures seem to be providing useful precedents.

Most of these reservations, however, are nitpicks. The really urgent need is to preserve the option for some form of mobile U.S. ICBM -- a crucial part of the restructuring of U.S. strategic forces to preserve their deterrent effect in a world of sharply reduced nuclear forces.

The START reductions in warhead numbers contribute to stability by reducing the number of attacking warheads aimed at strategic targets such as ICBM silos or shelters, bomber and tanker bases, command centers and the geographic areas in which ICBMs or ballistic missile submarines move. This makes it less likely that a first strike by either side could destroy the retaliatory capacity of the other side. We can reinforce this crisis stability unilaterally by increasing the number of strategic targets on the U.S. side. The best available way to do this -- and to protect against Soviet breakthroughs in anti-submarine warfare, unconstrained by treaty -- is to include in our force structure a mobile, highly accurate and quickly retargetable ICBM force component that can survive a large and accurate surprise attack with little or no warning.

Various basing modes and missiles can achieve this survivability. The small, single-warhead ICBM (Midgetman) in a hardened mobile mode, moving in peacetime over an 8000 square mile area on southwestern military bases and dispersing over a 20,000 square mile area on 20 minutes warning would be an ideal, though expensive, solution in an arms-control context; the Soviets would have to use 3000-7000 warheads to attack it.

To maximize the number of ballistic missile-armed submarines that the Soviets must track, we should stop building Trident subs, and turn to smaller ones, each with half the complement of D-5 missiles -- more expensive per warhead, but less vulnerable. We should also preserve the current limit of eight warheads per D-5 missile. Under the counting rules of SALT (almost sure to be carried over to START), testing the D-5 with twelve warheads would result in the U.S. having to forego one-third of its current number of submarine-launched ballistic missiles.

All of these desirable force structure improvements will be allowed by START -- if mobiles are allowed. So also will proceeding with the Advanced Technology Bomber and stealthy (low radar profile) Air-Launched Cruise Missiles to counter Soviet air defenses.

Finally, it must be clearly recognized that from the U.S. point of view, deep cuts in offensive forces could become very risky if Soviet ballistic missile defense were to be allowed to run free. In the arithmetic of deterrence, the stronger the defenses on either side, the greater the need for a sufficiency of weapons on the other side to overwhelm those defenses. Thus preservation of the ABM Treaty's ban on nationwide ballistic missile defense and, correspondingly, its limitations on testing and development of space-based defenses is an integral part of any long-term START limitations.

The worst possible outcome on ABM would be a situation in which the U.S. takes a course that ends in redefining or abrogating the ABM Treaty. This would allow the Soviets to indulge their historic proclivity toward defensive systems, causing us concern (even if ill-founded) about the capability of our retaliatory forces to penetrate them. Moreover, budgetary constraints and congressional limits would almost surely keep the U.S. from matching those efforts.

The current tendency on both U.S. and Soviet sides is to handle this issue by a verbal formula such as "agreeing not to withdraw from the ABM Treaty for seven (or ten) years," and allowing "as required, research, development and testing, which are permitted by the ABM treaty." This approach may be useful in getting through a summit, but it papers over a fundamental, and dangerous, disagreement.

The Soviets, the U.S. Congress, and the bulk of the non-governmental national security community interpret the ABM treaty and its duration in one way, the Reagan administration in another. All the Republican presidential candidates indicate, in varying degrees, favor for deploying SDI "when" "it" "works" (leaving all three of these terms undefined); the Democrats are for varying levels of research and, in some cases, technology development. The U.S. incoherence with respect to strategic defenses invites the Soviets to switch positions -- as they have on other issues in the past -- as their perceived advantage dictates. We should make sure that we and the Soviets will be playing by the same rules on SDI, and that confidence in deterrence will not be unexpectedly undermined by development and deployment of nationwide ballistic missile defenses.

Harold Brown is chairman of the Foreign Policy Institute at Johns Hopkins University and was secretary of defense in the Carter administration.