The maker of the first AIDS vaccine to be widely tested in humans said it will continue working for market approval of its product despite disappointing results announced yesterday from its large international clinical experiment.

In a population that included people from many racial and ethnic backgrounds, AIDSVAX was ineffective in preventing infection by the human immunodeficiency virus (HIV). There were hints, however, that it might be effective in black patients -- a scientifically puzzling and socially provocative finding that several experts characterized as dubious but company officials described as promising.

"We clearly will move toward licensure [of the vaccine]. The question is whether we do it with this study or with other ones," said Donald P. Francis, president of VaxGen, the California biotech company that conducted the study, using about 5,400 high-risk volunteers in the United States, Puerto Rico, Canada and the Netherlands.

It was unclear, however, on what grounds the vaccine might be approved in the foreseeable future. The vaccine offered no discernible protection to Caucasians or Hispanics, and the surprise findings that the vaccine was nearly 80 percent effective among black patients -- and nearly 70 percent effective among black, Asian and mixed-race patients combined -- were based on a subset of the total study population that was far smaller than customary for gaining approval by the Food and Drug Administration.

In all, there were about 500 non-whites in the study. Over the course of three years, 9.9 percent of those non-whites who received placebo (or inactive) injections became infected with HIV, compared to 3.7 percent who were given the active vaccine.

"Although it may be a meaningful result that would hold true when you look at larger numbers of African Americans, most professional statisticians would give you a large caveat about doing subset analyses on a relatively small group of people in a study that primarily showed a vaccine to be non-effective," said Anthony S. Fauci, chief of the National Institute of Allergy and Infectious Diseases. "My instinct tells me to be very careful about these results," said Fauci, who decided nearly a decade ago that the vaccine's formulation was not promising enough to warrant a large-scale federal trial.

Others yesterday decried what they called VaxGen's effort to "spin" the mostly negative results.

"Making any statements about efficacy in this subpopulation is grossly premature" because of the small number of people involved, said Gregg Gonsalves, director of treatment and prevention advocacy at Gay Men's Health Crisis in New York. Gonsalves urged the company "not to overstate the promise of its product based on scant and inconclusive data."

John Moore, an AIDS immunologist at Weill Cornell Medical College in New York, was similarly blunt, saying he knew of no biological mechanism that could explain the apparent difference in response between blacks and whites.

"Antibody responses are not racially driven, as far as I've seen," he said. "I don't think this is real. It's just one of those statistical flukes."

AIDSVAX consists of a genetically engineered copy of part of the outer coat of the AIDS virus, a structure called gp120. The molecule prompts the body to produce antibodies, but not all such antibodies can neutralize HIV. Some scientists suspect that additional immunological responses may also have to be provoked if a vaccine is to be effective.

VaxGen researchers said they were intrigued by an apparent link between higher levels of neutralizing antibodies and lower rates of infection seen in the blacks in the study. They are also trying to identify the precise structure of AIDS-protecting antibodies, which in turn could inform future efforts to develop an effective vaccine.

Among the many problems with AIDSVAX is its dosing schedule, at least as tested in the recent study. Volunteers were vaccinated at the start, at one month, at six months and then every six months, with many volunteers receiving seven shots. None of the immunizations now endorsed for widespread use by the World Health Organization requires more than three injections.

The study results were based on the 5,009 people who received at least three shots, out of 5,417 participants originally enrolled. The vast majority were gay and bisexual men, along with about 300 women at high risk of acquiring HIV from their male sex partners. A similar study by the company is continuing in Thailand, with results expected this summer.

VaxGen shares dropped precipitously following yesterday morning's announcement, closing at $6.86, a 47 percent decline for the day.

Staff writer Justin Gillis contributed to this report.

Donald P. Francis, president of VaxGen, the California biotech company that produced the first AIDS vaccine to be widely tested in humans, speaks to reporters about the results of a large international study.