Deputy White House press secretary Larry Speakes said yesterday that President Reagan had made some "preliminary decisions" on the controversial MX missile -- thought to be a compromise, cut-down version of the multiple-shelter, land-based system.

Under the package being discussed in the Pentagon and on Capitol Hill, the giant intercontinental ballistic missiles would be deployed in two stages.

As an initial step, the first 100 MX missiles would be rotated among 1,000 shelters that would be built, beginning in 1986, on public land primarily in Clark County, Nev., near Las Vegas. That is one of the few areas in the West that said it wanted the new ICBMs.

While this land-based deployment program was under way, the Pentagon would accelerate research on several other systems to provide basing for additional MX missiles, or provide an antiballistic missile defense for the first 100 in an effort to make them invulnerable to a Soviet attack.

Among other systems to be explored in the next few years under this plan, sources said, are the new air-mobile concept built around the so-called "Big Bird" long-flying airplane, deep underground shelters that would provide for a secure reserve of missiles, and putting more shelters around the 100 MX missiles.

The plan for starting off with 100 MX missiles and 1,000 shelters was one of the alternatives discussed last Monday at a special National Security Council meeting devoted to the strategic weapons decisions facing the president. As a reduced version of the Carter administration's plan of rotating 200 missiles among 4,600 shelters, sources said, it should create fewer political and environmental problems and thus should be more acceptable to the president.

Speakes told reporters that Reagan "knows the direction he wants to go" on the MX and directed Defense Secretary Caspar W. Weinberger "to proceed in certain directions."

He would not give any more specifics. At the Pentagon yesterday, aides to Weinberger said he would not comment "on his private discussions" with Reagan.

Speakes said Reagan would not announce his final decision until September, and no one was willing yesterday to say that the compromise land-based shelter idea was the one he had chosen.

Speakes indicated that the initial basing favorite of both Weinberger and the president, a new concept for putting the MX aboard airplanes, had faded as the initial deployment idea. Speakes conveyed that impression by saying that he "would not dispute" the statement by Sen. John G. Tower (R-Tex.) on Friday that the air-mobile idea was no longer under serious consideration.

Reagan, during the 1980 campaign and after taking office, has said he was opposed to any basing for the MX that would damage the environment.

But recently, White House aides have pointed out that such statements by Reagan have always been coupled with the idea that if there were no other alternative and national security required it, he would approve some land-basing plan.

Another idea that had been raised by Weinberger at the NSC meeting last Monday was turning away from the MX and substituting a missile that would be used by both the Air Force and the Navy.

Sources said yesterday that the common-missile idea had been attacked inside and outside the Pentagon, particularly by legislators on Capitol Hill, including Tower.

Weinberger is scheduled to meet with the president tomorrow in California, but the subject is described as the defense budget. David A. Stockman, director of the Office of Management and Budget, is to attend that session.

The need for deceptively basing the MX developed out of fear that the Soviets, with the lead in nuclear warheads, could wipe out the entire U.S. land-based ICBM force, including the new MXs, unless they were hidden among so many targets that they drained off all the Soviet missiles.

Supporters of the notion for starting out with 100 MX missiles to be hidden among 1,000 shelters say the plan has its roots in a recommendation by the Townes Commission, a group appointed in February by Weinberger to come up with an MX basing plan that was less costly in dollars and political opposition than was the Carter plan.

A majority of that group, made up primarily of scientists and defense experts, supported the two-stage approach, taking some preliminary steps while researching others.

The Townes suggestion was to put 100 MX missiles in 100 shelters, but place them so that additional shelters could be added, allowing the missiles to be moved between shelters so that they would be harder for Soviet missiles to hit.

The Air Force and others developed the notion that it was better to begin by hiding the initial 100 missiles among 1,000 shelters.

With just that many shelters, they also were able to place the entire system in one valley covering two counties -- avoiding, they hope, many of the political problems raised by the larger system, which covered two states and 12 counties.