The Food and Drug Administration yesterday prepared to wield, for the first time, a drastic power Congress gave it 15 years ago: ruling a medicine an "imminent hazard" to public health and halting sales outright.

The drug is phenformin, sold by Ciba-Geigy Corp. as DBI and by USV Pharmaceutical Corp. as Meltrol. Doctors prescribe it to lower blood sugar in at least 250,000 adults with mild or so-called onset diabetes.

If the FDA follows normal procedures in trying to stop sales of phenformin, the legal process could last two years or so. Meanwhile, sales - now running, at retail, at about $25 million annually - could continue unimpeded.

The FDA had not used the imminent hazard power even in extreme cases, including an antibiotic product called Panalba, which, by the agency's own estimate annually was causing hundreds of thousands of needless injurids, a few of them fatal. After losing a court battle with the FDA, the Upjohn Co. stopped selling Paanlba in 1970.

In the past the FDA has defined imminent hazard so strictly as to preclude its use. Recently, however, the courts have begun to define "imminent hazard" in ways that seem to fit phenformin.

FDA lawyers note, for example, that last November the U.S. Court of Appeals here held that a pesticide can be ruled an imminent hazard if there is a mere "substantial likelihood" that it could cause serious harm if sold duing the year or two an administrative proceeding was under way.

In the phenformin case, and FDA panel of outside experts last October recommended unanimously that sales he halted because of a "clear and unequivocal association" between the pills and lactic acidosis, an adverse reaction fatal in half those in whom it occurs. In the seven months since then, the FDA has received reports of 22 fatal and 24 nonfatal cases in phenformin users.

But yesterday, at an unprecedented - and sharply divided - hearing called by the FDA as a prelude to the expected sales cutoff - scientists testified that reports to the agency grossly understate the true toll.

Former panel chairman Thaddeus E. Prout of Johns Hopkins University School of Medicine, urging a ban in behalf of the American Diabetes Association, termed it "undoubtedly true" that the FDA never hears of a large share of lactic acidosis cases.

Emphasizing that on one has demonstrated phenformin to be a life-saving drug, Prout, speaking for himself, said of a ban "the sooner, the better."

Citing evidence from various hospitals and sources around the world, including a medical officer in the FDA, Dr. Sidney M. Wolfe estimated that phenformin needlessly kills at least 1,000 user a year. At the same time, he said, "There is no evidence of any death occuring as a result of the lack of availability of phenformin."

Wolfe, director of the Health Research Group, which is financed by Ralph Nader's Public Citizen, Ind., petitioned Joseph A. Califano Jr., Secretary of Health, Education and Welfare, on April 21 to rule phenformin an imminent hazard.

At Califano's request, the FDA then set yesterday's hearing, which was conducted by Dr. J. Richard Krout, the agency's drugs chief. He told reporters he is "considering very seriously" recommending a ban to Commissioner Donald Kennedy.

Wolfe said the HRG will go to court to seek a ban if "this lethal drug" is not ruled an imminent hazard next week.

Opposing scientists and a Ciba-Geigy vice president contended that "chaos" and "panic" would occur among phenformin users if the drug is abruptly withdrawn. They said that if used in accord with strict new prescribing instructions adopted by the FDA last January phenformin poses a aminimal hazard that is outweighed for some patients, particularly obese persons in whom chemically different sugar-lowering pills don't suffice.