Among the worst horrors of World War II were the Nazi medical experiments conducted on prisoners in the death camps. In one series of experiments, victims were forced into tanks of ice water and studied as they froze to death. The ostensible purpose was to gather information that might prevent the deaths of German pilots shot down over the North Sea.

This happened a long time ago in what we hope was a uniquely evil society. Nevertheless, we still have concerns about the use of human subjects in medical research. People speak of being "guinea pigs" in experiments, and some fear that their well-being may take second place to the success of the research project. Do we still need to worry about the rights of human subjects in medical research? What are the safeguards today?

The fall of the Third Reich certainly marked the end of the systematic maiming and killing of helpless people by the state in the name of medical research. The Nuremberg Code of 1947, an explicit response to the Nazi experiments, spells out the rights of research subjects in a way that leaves no doubt that their welfare must be held paramount. The Nuremberg Code was followed by a number of similar formulations, including the World Medical Association's Declaration of Helsinki, first issued in 1965, and, in this country, the 1983 federal regulations for the protection of human subjects.

All have in common two essential provisions. First, human subjects of medical research must understand all relevant information about the experiment and give their voluntary consent. Second, they cannot be subjected to disproportionate risks, even with their consent, since some patients might be willing to do nearly anything to please their doctors. What's more, federal regulations require medical schools and hospitals to set up committees -- institutional review boards -- to evaluate the ethical aspects of proposed research.

These safeguards are necessary. Henry Beecher, in a widely publicized 1966 article in the New England Journal of Medicine, described American experiments in which human rights were flagrantly disregarded. An example was the Willowbrook hepatitis study, in which hepatitis was studied by deliberately infecting retarded children with the virus. This was done with the permission of the parents.

Equally notorious is the Tuskegee syphilis study, which involved withholding treatment from men with syphilis to observe the course of the disease. The study, conducted by researchers from the U.S. Public Health Service among blacks in rural Alabama without their informed consent, lasted from 1932 to 1969, long after the Nazi crimes were revealed.

Fortunately, such incidents have become rare as our medical research establishment over the past 20 years has more explicitly come to grips with the issue of unethical research. Still, some unethical research continues today. Usually, I believe, it stems from the desire to get unambiguous answers to scientific questions as quickly as possible. In comparison with the Nazi experiments, the violations are minor and often arguable, but we still need to do what we can to identify them.

What sorts of violations am I talking about? A relatively frequent one is including subjects in an experiment without making sure they understand the way the study will be carried out. For example, researchers may be reluctant to tell potential subjects that treatment will be determined randomly, by something like a coin toss, because the subjects might refuse to participate. A researcher may therefore gloss over this aspect of the experiment. The fact that it is spelled out in written form does not absolve the researcher of the responsibility to make sure it is understood.

Another frequent violation is comparing two treatments when there is good reason to believe one is better than the other. This exposes those who receive the inferior treatment to unnecessary risks. Occasionally, researchers do it anyway because such a comparison enables them to gauge just how much better a new treatment is.

An important aspect of this issue is whether the results of unethically conducted experiments should be published in medical journals. This question was intensely debated with regard to the Nazi freezing experiments for nearly 50 years. Those who opposed publishing the results felt that doing so would dishonor the victims and tend to legitimize unethical research. In contrast, those who favored publishing the information believed that we should redeem whatever small benefits we could from the atrocities. Why not use the knowledge gained in the Nazi experiments, at the cost of such suffering, in an attempt to prevent future suffering?

Recently, a paper published in the New England Journal of Medicine analyzed the validity of the freezing experiments and found them scientifically worthless, as well as inhuman. The broader and more immediate question, however, remains: should results of current research in which rights of human subjects were not fully honored be published?

We at the New England Journal of Medicine have decided not to publish reports of unethical research, regardless of scientific merit. We believe that if all journals had such a policy and were sensitive to subtle violations of ethical standards, it would be a strong deterrent to unethical research. Publication is an important part of the reward system in medical research, and researchers would exert care not to cut ethical corners if they knew this would jeopardize publication.

The rights of human subjects should take precedence over the usefulness of any new information gleaned in research. Our quest for scientific knowledge -- important as it is -- must not erode our commitment to human rights.Marcia Angell, a physician, is executive editor of the New England Journal of Medicine. Second Opinion is a forum for points of view on health policy issues.