In September, 1980, the Environmental Protection Agency proposed a seemingly modest rule to deal with the possibility that some children exposed to crumbling asbestos in more than 8,600 public and private schools will die from lung diseases 20 years after inhaling tiny fibers of the insulating material.

The agency recognized that existing scientific studies do not allow officials to make a precise estimate of the risks involved. But, it said, they do allow a crude estimate: over a 30-year period, 100 to 7,000 of the 3 million pupils and 250,000 staff members in the affected schools would die prematurely.

The numbers and the way they were figured drew heavy fire from the industry and also from independent scientists. So EPA took another look and last July reduced the outside number of possible victims to 3,000.

Now, however, the EPA has deleted the numbers entirely from the document supporting the proposed rule, which would require local education agencies to inspect public and private school buildings to for crumbling materials that contain asbestos and notify parent-teacher associations and school employes, who account for an estimated 10 percent of the possible victims.

The deletion pleases the Asbestos Information Association (AIA), which had pushed for it on behalf of its more than 50 member companies. But it infuriates Rep. George Miller (D-Calif.), chairman of the House subcommitttee on labor standards, as well as Dr. Winona Vernberg, chairman of an EPA panel of outside scientific advisers that EPA officials did not consult when they took the action. It may fuel charges that EPA, under the Reagan administration, has taken on a pro-industry tilt.

Without the numbers, which were reached in "a good faith effort" by EPA scientists and advisers, Miller said, the "sense of urgency" in implementing the rule is lost for local officials, parents and school employes.

Citing evidence that EPA officials consulted only the AIA before making the deletion, he said, "I am concerned about adverse pressure on EPA by the asbestos industry to circumvent the rule-making process."

Vernberg, dean of the University of South Carolina School of Public Health, said that the deletion and the way her panel was bypassed is "destroying any credibility at all that EPA would have and is ignoring public health consequences."

Vernberg heads the Toxic Substances Subcommittee, a unit of the EPA Science Advisory Board. The agency's Office of Toxic Substances hasn't convened the advisory subcommittee once this year; previously, it had been calling the panel four to six times a year. "Why have a subcommittee if you're not going to listen to it at all?" she asked.

According to an internal EPA memo, Don R. Clay, EPA deputy assistant administrator for toxic substances, and the agency's general counsel's office suggested that the estimates be deleted. Both Clay and Timothy S. Hardy, a Washington attorney for the AIA, which has not opposed the proposed rule itself, defended the deletion.

Clay emphasized that the rule, originally intended to be made final last February but still pending, is intact, and "does not need a detailed, quantified risk assessment to support it because it would cost very little" to implement.

The deletion was "a judgment call" that avoids useless controversy and will "help get the rule out," he said. The idea for taking out the numbers came from his staff, with the AIA exerting "negligible" influence, he added. He was unable to recall any role played by the general counsel's office.

Hardy says that there is a question whether the low-level asbestos exposure at issue is a hazard at all, that the deleted numbers were "poorly documented," and that their inclusion "could in an unwarranted way alarm the public."

In a Oct. 22 meeting with two aides to Rep. Miller, Clay said he was deferring the risk assessment for possible inclusion in a forthcoming proposed rule that would require the school districts to take corrective action. Clay later told a reporter he hadn't realized that EPA had abandoned the rule with a public notice in April, four months before he joined the agency.