Top members of the House and Senate intelligence committees are carrying on a behind-the-scenes debate over the nation's spy satellite system, with one group arguing that the government could save hundreds of millions of dollars by building a new generation of small satellites to be launched in place of some of the big ones already planned.

The new chairman of the House Permanent Select Committee on Intelligence, Rep. Larry Combest (R-Tex.), has put together a group of House and Senate members that wants to reduce the number of big, new 30,000-pound satellites, each costing about $1.5 billion, whose launchings are scheduled to begin in 1998.

The Combest group wants to start production now of 2,000-pound space vehicles, each costing about $100 million and nicknamed "small-sats." One advantage of the smaller platforms is that they could be launched quickly during a crisis to supplement the work of the big satellites in space.

Opposing Combest is Rep. Norman D. Dicks (Wash.), ranking Democrat on the House intelligence committee, who, along with other House and Senate intelligence panel members, wants the National Reconnaissance Office (NRO) to continue current plans to launch the big satellites, beginning with one known as the 8X.

Sen. Arlen Specter (R-Pa.), chairman of the Senate Select Committee on Intelligence, said members of his panel "were on both sides of the issue."

The debate is about $10 billion in future spending on U.S. spy satellites, which each day generate a flow of thousands of intelligence photos and radar images from space. Still cloaked in Cold War secrecy, the U.S. space imagery program has cost $100 billion over 30 years but has long been among the country's most successful intelligence programs, according to active and retired CIA officials.

Although the interception from space of private telephone and electronic messages has also been impressive, it has always been the dazzling detailed photo coverage from hundreds of miles in space that has had presidents, policymakers and military leaders regularly asking for more.

The dispute in Congress has arisen as the NRO already has drawn criticism recently for accumulating a pool of $1.6 billion of unspent funds without notifying its supervisors at the Pentagon or CIA, or its congressional overseers. More than $1 billion of those unspent funds was recently transferred to other Pentagon and intelligence programs.

The debate is taking place as the space satellites' basic purpose is changing in the Cold War's wake. CIA Director John M. Deutch recently said that satellite imagery -- which began as a tool for gathering information about Soviet strategic bombers and missiles, and then became the backbone of arms control verification -- "now {has as} the big user, the military commander" on the battlefield.

In Cold War days, the program was unconstrained by dollar concerns, Deutch said. Today that too is changing.

The United States is three years into a 6- to 10-year imagery plan that could be less than the $10 billion agreed to in the last year of the Bush administration. It includes development and production of four or five of the big, 8X satellites, carrying both radar and electro-optical cameras. A new element on some of the new satellites will provide broader coverage of any landmass, including future battlefields. This fix would answer complaints voiced during the 1991 Persian Gulf War by Gen. Norman H. Schwartzkopf that he was not getting a wide enough picture.

The Cold War satellites looked down "like soda straws," one former intelligence analyst said recently, but "the 8X is like eight straws put together."

Two years ago, the then-chairman of the Senate intelligence committee, now retired Dennis DeConcini (D-Ariz.), sharply opposed the 8X and tried to cut its funding. He argued in closed session, a colleague said, that the current imagery satellites "could last until at least 2010. The U.S. should take time before pushing ahead with the jerry-built 8X that couldn't be launched until 1998-2000 time frame at a cost of a couple of billion dollars."

DeConcini urged, the colleague said, that "we try for something new and be ahead of the curve."

NRO believed that with DeConcini gone the proposed 8X satellite program was set. However, this year the debate took a different tack when Combest and others took up where the Arizonan had left off.

The new small-sats, which are under development, would have shorter lives in space than the 8X but would provide more flexibility in coverage, according to supporters.

A former senior NRO official said some small satellites were built earlier to serve as quick replacements for the larger ones. However, he said this was the first time a push was being made to make small-sats part of the program.

"They may not end up as inexpensive as they seem," he added, "since they may need special crews on call to launch them in a crisis."

Combest and other members refuse to discuss the matter on the record. But one member of the Combest group recently criticized the NRO for "tunnel vision and a bias against ideas that threaten the existing program they want to follow."

The Combest committee's report on the intelligence authorization bill for this fiscal year reflects the chairman's view that "over the past 10 years, there have been major technology advances that reduce spacecraft weight, volume and power requirements. Since launch has been a primary cost driver," the report goes on, "these potential weight reductions, coupled with new launch options, present the possibility of substantial savings even while largely retaining or even increasing spacecraft capabilities." "The NRO must learn to balance technical elegance with cost-efficient solutions," the report says.

During the Sept. 13 debate on the intelligence authorization bill, Dicks said he was worried that the stability of the currently planned satellite mix could be harmed by rushing off to try new, untested systems.

"We should slow down on small-sats,' " a member supporting Dicks said. "There is no rush to go into production. The {space satellite} architecture we have now is solid." Today, orbiting 200 to 300 miles in space, the United States has three huge operating KH-11 and KH-12 "Keyhole" photo-electronic satellites that send data down to ground stations here and abroad that instantly are turned into photos so detailed that they can see things as small as an inch. There are also two giant "Lacrosse" radar-imaging space vehicles that can see through clouds and at night with resolutions that approach photography.

In addition there are newer versions of the Keyhole and Lacrosse space vehicles in sanitized "clean rooms" in California and elsewhere. They have not yet been put into orbit because the older systems are working far longer than originally expected.