The United States conducted at least 12 and perhaps as many as 19 unannounced nuclear-weapon tests in the Nevada desert from 1982 through 1984, according to a private research group that based its estimates on seismic data and other sources.

The calculations by the Natural Resources Defense Council provide the first indication of the scope of U.S. testing since the Reagan administration began conducting some tests secretly, apparently in 1982. During those three years, the Energy Department exploded another 44 devices in announced tests.

The pace of the U.S. testing program, apparently about one-third more active under President Reagan than during the Carter administration, has become a greater issue since last summer, when Moscow declared a unilateral testing moratorium and urged Washington to follow suit. Many U.S. scientists have said a test ban is the best way to slow the arms race.

The administration has rejected the Soviet overtures, arguing that nuclear tests are essential to develop new weapons and maintain confidence in older ones, and that compliance with a test ban would be difficult to verify. Officials also defended secret testing as bolstering national security by shielding information from the Soviets.

Some of the unannounced explosions revealed in today's report were detected and reported, without explanation, by the U.S. Geological Survey or Sweden's Hagfors Observatory, which monitor earthquake and other ground-shaking activity. But several of the unannounced tests were not detected by either institution, a lapse that raises questions about how effectively a test-ban treaty could be policed.

Three of the Defense Council study's authors -- Thomas B. Cochran, Robert S. Norris and William M. Arkin -- criticized the secrecy in an interview this week and said their report shows that all but the smallest tests can be detected. The fourth author, Milton M. Hoenig, was not present.

The council is a nonprofit group that concentrates on environmental issues but has also studied nuclear proliferation and other matters relating to nuclear weapons.

Arkin said the unannounced tests stand in contrast to the administration's public-relations policy for its Strategic Defense Initiative, the "Star Wars" research program aimed at defending against nuclear missiles. In an effort to build support for the SDI, the Defense Department frequently issues film and statements about successful tests of defensive weapons, he said.

"Every aspect of every kooky idea in SDI is trumpeted," Arkin said. "But they don't want people to think about the fact that they are also working on the next generation of nuclear weapons."

Arkin also said the public should know what types of "new generation" nuclear weapons, such as atomic antisatellite weapons or small "backpack" atomic mines, are being developed.

Chris L. West, an Energy Department spokesman in Las Vegas, said the government announces all large explosions but conceals smaller ones that the Soviets might not detect.

"The rationale is, why hand our adversaries national security information for the price of buying a newspaper?" he said yesterday. "There's a low level of test that's extremely difficult, if not impossible, to distinguish from an earthquake."

In addition, U.S. officials noted that the Soviets never announce tests. Experts said the U.S. military can detect all but the smallest explosions through monitoring facilities in Norway, Turkey, China and other nations.

From 1982 through 1984, the Soviets conducted at least 32 nuclear tests, according to Energy Department spokesman David S. Devane, who cautioned that some tests may not have been detected and said others are not announced by Washington "to protect intelligence sources."

Since 1963, when atmospheric tests were banned, U.S. policy has called for announcing some underground explosions and concealing others, West said. But he also said that, from 1975 until early in the Reagan administration, virtually all tests were announced anyway.

"I think there's been a rebirth of security consciousness," he said.

The authors of today's report said they relied on official documents, including Geological Survey reports and a paper presented last summer by a Lawrence Livermore National Laboratory scientist, Ray E. Kidder.