Television and newspapers are full of stories about a new study showing that women who take estrogen after menopause have an increased risk of breast cancer. It has been known for some time that such women have a lower risk of heart disease and osteoporosis -- thinning of the bones. So what should postmenopausal women do? Indeed, what can any of us do about all the health risks we learn about from the media?

Nearly every week, a new study is said to show that something is good or bad for your health and millions of Americans try to incorporate that information into their already complicated rules for living. Exercise regularly, avoid stress, eat bran and fish oil and Vitamin A, avoid saturated fats. Sometimes it seems impossible to get it right, as discovered by those who gave up regular coffee because of its stimulant effects, only to learn that decaffeinated coffee may increase cholesterol.

No one, of course, could possibly follow all this advice. There is just too much of it, and besides, some of it is contradictory -- as in the case of postmenopausal estrogens. So what are the consumers of all this information to do? Surely not all of the health news is equally important or valid. Is there any way to sort out which should be acted on and which may be safely ignored?

Here are a few guidelines.

First, ask whether what you are advised to do, or to avoid, is thought to have a big effect or a small one on the risk of getting the disease in question. For example, does cigarette smoking have a big or small effect on the risk of getting lung cancer? Most advice is based on studies showing that a disease, such as lung cancer, is more common among people who live a certain way -- in this case, among cigarette smokers. Smoking is then termed a risk factor for lung cancer because it increases the risk of getting it. The question is, by how much? In this case, the increase is enormous; smokers are many times more likely to get lung cancer than are nonsmokers.

In contrast, cholesterol has a smaller effect on the risk of heart attacks; men with high cholesterol levels are about twice as likely to have a heart attack as are those with low cholesterol levels. Other risk factors have very small effects; for example, the risk of breast cancer in women who drink alcohol is only 1.3 times that of women who don't -- about the same as the new study found in women who take postmenopausal estrogen compared with those who don't.

Second, ask how common the disease is that the advice is meant to prevent. If it is very rare, even an important risk factor may not present much of a threat. Cancer of the esophagus, for instance, is relatively rare. Although smokers are many times more likely to get it than are nonsmokers, it is not common even in smokers. For this reason, the statement that something doubles the risk of a disease tells us very little about the danger unless we know how great the risk was in the first place.

And third, ask how strong the evidence is on which the advice is based. If the advice is based on a single scientific study -- even one reported in the New England Journal of Medicine -- wait. Very few scientific studies are conclusive enough to warrant changing behavior on the basis of the results. Instead, wait to see whether other studies confirm the new findings.

You should also be aware of how difficult it is to be sure that the connection between the risk factor and the risk is direct, rather than indirect or coincidental. For example, children of smoking mothers have more respiratory disease than other children do, but it is possible that this connection is indirect. Perhaps mothers who choose to smoke are also more casual about exposing their children to respiratory infections or about treating them promptly.

For these reasons, we should not automatically respond to every media report that medical research has shown a new risk in daily living. Often, what has been shown is a small increase in a small risk, based on inconclusive evidence. Why then do researchers, medical journals and the media give so much attention to such findings?

One reason is that risk factors of almost no significance to individuals because their effects are so small -- such as alcohol as a risk factor for breast cancer -- may be important as a public health problem to the population as a whole. This is especially true if the disease is common.

An analogy is the importance of car accidents to the public health, even though the risk for each of us may not be enough to deter us from driving. Similarly, although the probability of an individual woman preventing breast cancer by not drinking is so small as to be not worth considering, if no American women drank we might have significantly fewer cases of breast cancer per year. This line of reasoning is usually used to demonstrate cost savings to society by avoiding some risk factor.

So what should we as individuals do about all the advice? First, if it concerns something that has repeatedly and consistently been shown to increase greatly the risk of a fairly common disease, pay attention. An example is the link between cigarette smoking and lung cancer.

At the other extreme, one may wish to ignore any advice to curtail an ordinary, pleasurable activity that increases a small risk slightly, particularly if the connection is based on one study. So go ahead and have another cup of coffee, even though it slightly increases the risk of bladder cancer.

For intermediate situations -- those involving a small increase in the risk of a common disease or a large increase in the risk of a rare disease -- use common sense. This means moderation, both in what you do and what you avoid.

Marcia Angell, a physician, is executive editor of the New England Journal of Medicine.