WASHINGTON (AP) _ A consumer health group accused Eli Lilly and Co. of covering up a problem with the experimental hepatitis drug that killed five patients in a National Institutes of Health trial last summer.

Public Citizen's Health Research Group said Thursday the drug company should have given the Food and Drug Administration earlier notice that a healthy volunteer in a separate trial showed signs of liver toxicity last March after taking just two doses of the drug, fialuridine (FIAU).

The company denied the allegations.

In a letter to FDA Commissioner David Kessler, the Health Research Group said three of the five patients who died began taking FIAU ''after the normal volunteer was hospitalized in Indianapolis.''

''There is no question that Eli Lilly should have stopped the NIH trial by the end of March or in the first days of April,'' it said. The FDA might have halted the experiment ''had Lilly not covered up this important data.''

Edward A. West, Lilly's director of corporate communications in Indianapolis, Ind., called the assertion ''that we covered up data ... irresponsible and inaccurate.''

''It is not borne out by the facts in any way,'' said West, who added it ''was impossible to predict the toxicity that was seen in the NIH trials.''

Five of the 15 patients with the hepatitis B virus died in the NIH experiment; two others survived only after getting liver transplants.

A FDA task force concluded last month that researchers in earlier experiments failed to spot early warning signs, including a possible link to five other deaths.

The FDA has ordered tighter scrutiny of clinical trials, and it is reviewing whether the FIAU researchers, Lilly and other drug companies violated any rules.

The Health Research Group's Sidney M. Wolfe and Robert Reid, in a nine-page critique of the FDA task force report, disputed the FDA's conclusion that there was no evidence of toxicity in animal studies on FIAU.

They said Lilly should have notified the FDA within 10 days that a healthy volunteer had been hospitalized in March 1993; instead the company waited for four months.

West said the volunteer showed no symptoms of illness but was withdrawn from the study because he exhibited ''mild elevations in liver enzymes which may be seen in many medications.'' The experiment on healthy volunteers was intended to compare liquid and tablet forms of FIAU.

He said the volunteer stayed on by choice in an Eli Lilly clinic for observation partly to help him avoid alcohol. The man spent 41 days in the clinic, 13 more than he signed up for, West said.

West said Lilly followed all of the FDA's regulations and reported the event ''just as we should have'' in an annual report filed in August.

Dr. Roger Williams, who headed the FDA task force, said it was impossible to say whether the lethal trial would have been halted sooner if the FDA had known about the healthy volunteer's problem.

''The data would have come in after the start of the NIH study,'' said Williams. ''Would that have halted further investigation if FDA had known about it? I don't know.''

The experiment was halted in July after two patients died.