The company, DeCode Genetics, first found the variant gene among Icelanders and then looked for it in three American populations, in Philadelphia, Cleveland and Atlanta.

Among Americans of European ancestry, the variant is quite common, but it causes only a small increase in risk, about 16 percent.

The opposite is true among African-Americans. Only 6 percent of African-Americans have inherited the variant gene, but they are 3.5 times as likely to suffer a heart attack as those who carry the normal version of the gene, a team of DeCode scientists led by Dr. Anna Helgadottir reported in an article released online yesterday by Nature Genetics.

Dr. Kari Stefansson, the company's chief executive, said he would consult with the Association of Black Cardiologists and others as to whether to test a new heart attack drug specifically in a population of African-Americans.

The drug, known now as DG031, inhibits a different but closely related gene and is about to be put into Phase 3 trials, the last stage before a maker seeks the Food and Drug Administration's approval.

Last year a drug called BiDil evoked mixed reactions after it was shown to sharply reduce heart attacks among African-Americans, first in a general study and then in a targeted study, after it failed to show efficacy in the general population. The drug, invented by Dr. Jay N. Cohn, a cardiologist at the University of Minnesota, prompted objections that race-based medicine was the wrong approach.

Geneticists agree that the medically important issue is not race itself but the genes that predispose a person to disease. But it may often be useful for physicians to take race into account because the predisposing genes for many diseases follow racial patterns.

The new variant found by DeCode Genetics is a more active version of a gene that helps govern the body's inflammatory response to infection. Called leukotriene A4 hydrolase, the gene is involved in the synthesis of leukotrienes, agents that maintain a state of inflammation.

Dr. Stefansson said he believed that the more active version of this gene might have risen to prominence in Europeans and Asians because it conferred extra protection against infectious disease.

Along with the protection would have come a higher risk of heart attack because plaques that build up in the walls of the arteries could become inflamed and rupture. But because the active version of the gene started to be favored long ago, Europeans and Asians have had time to develop genetic changes that offset the extra risk of heart attack.

The active version of the inflammatory gene would have passed from Europeans into African-Americans only a few generations ago, too short a time for development of genes that protect against heart attack, Dr. Stefansson suggested.

The DG031 drug being tested by DeCode Genetics affects a second gene, but one that is also involved in control of leukotrienes. Because the drug reduces leukotriene levels and inflammation, it may help African-Americans who have the variant of the hydrolase gene. "It would make scientific, economic and particularly political sense to have a significant part of the clinical trials done in an African-American population," Dr. Stefansson said.

A spokeswoman for the black cardiologists' group, which supported the BiDil trial, said the group's officials were not ready to discuss the new gene.

Dr. Troy Duster of New York University, an adviser to the federal Human Genome Project and a past president of the American Sociological Association, said he saw no objection to a trial, provided it focused on African-Americans with the risk-associated variant of the gene and took into account that people with ancestry from different regions of Africa might show variations in risk.

But Dr. Charles Rotimi, a genetic epidemiologist at Howard University, said a separate study of African-Americans would not be desirable. The variant gene may be overactive in African-Americans because of their greater exposure to deleterious environments, Dr. Rotimi said.

Dr. Cohn, the inventor of BiDil, said it was "always best to study a drug in a highly responsive group," rather than testing large populations where possible benefits to subgroups could be missed.