Every day, millions of Americans are bombarded with health information through direct-to-consumer drug ads on TV. Millions more go online to research a new diagnosis, search for treatments, or simply cruise Web sites devoted to health.

In the month of September alone, 51.9 million Americans – or 33% of all people who use the Internet – visited health-related Web sites, according to Com-Score Media Matrix, a Reston, Va.-based research firm that tracks Internet use.

The trouble is, much of the information thrown at us on TV or offered up on the Internet is misleading or downright wrong. Even reliable-looking research articles can be less meaningful than they appear. How can you tell the difference between credible information and half-truths? Here are 10 tips:

1. When cruising the Web, take the time to notice who wrote what you’re reading. It’s easy to just keep clicking on links and find yourself sliding from a reputable site, like the one on cancer run by the federal government (www.nci.nih.gov), to sites run by pharmaceutical companies or advocacy groups with their own agendas. A group called the Health on the Net Foundation (www.hon.ch) scrutinizes sites for accuracy and offers a seal of approval and links to approved sites.

2. Check the attribution of any medical article you stumble upon. Usually, the most reliable studies are those published in an authoritative medical journal such as the Journal of the American Medical Association (http://jama.ama-assn.org) or the New England Journal of Medicine (www.nejm.org), or other journals whose articles have been reviewed by experts. While you’re at it, look to see who funded the study – if it’s the company that makes the drug, be a bit more skeptical of the results.

3. Get a sense of how thorough a study is and exactly what questions it’s asking. Some studies only involve a handful of people, others many thousands. Some studies are “observational,” meaning that patients just took a drug and doctors tracked what happened. Others are “randomized, controlled double-blind clinical trials,” in which patients are chosen by chance to get, say, a new drug or a placebo, and neither they nor their doctors know at the outset who is in which group. Be sure to notice if a drug was tested on people or rodents. Mice can react very differently to drugs than people.

4. Pay attention to how much benefit patients really got. A new cancer drug may extend life, but it’s important to know if it extends life for two months or two years.

5. Figure out whether an ad or a study is talking about getting a disease or dying from it. The oft-cited, and true, figure that one in eight women will get breast cancer over a lifetime is very scary. Certainly, no one wants to be diagnosed with cancer, but only one in four women who gets breast cancer actually dies of it, said Dr. Eric Winer, director of the breast oncology center at the Dana-Farber Cancer Institute. In other words, three in four women who get breast cancer die of something else.

6. Be wary of studies that lump together several outcomes, what researchers called “combined end points.” One ad touting the cholesterol-lowering drug Lipitor lumps together fatal and nonfatal heart attacks, saying that Lipitor reduces these outcomes by 36%. But the fine print says the reduction in fatal heart attacks, taken separately, was not statistically significant, which means it could have happened by chance, said Dr. Lisa Schwartz and Dr. Steven Woloshin, researchers at the Veterans Administration Outcomes Group and Dartmouth Medical School.

7. Be on the lookout for apples-to-oranges comparisons. The 5-year survival rate data makes it look as though there’s been a huge improvement in the prostate cancer survival rate over the last 30 years. In 1974, 33% of all men diagnosed with the disease died within five years; from 1995-2000, fewer than 1% of those diagnosed died. But the actual mortality rate barely budged over that time. The reason the 5-year survival rate is so impressive is that we’re catching prostate cancer much earlier now, so men are living longer with the knowledge that they have the disease. And many men whose cancers will never turn deadly are now getting diagnosed with the disease, Drs. Schwartz and Woloshin said.

8. Before dumping prescriptions or changing behavior, make sure to understand how big the risk really is. Many doctors and postmenopausal women may have overreacted two years ago when a link was found between hormone replacement therapy and breast cancer – and millions abruptly stopped taking their daily pills. A large study showed that women taking hormone replacement therapy had a 26% increase in the risk of breast cancer. A less frightening way to state the same thing is to note that 38 out of 10,000 women taking hormones got breast cancer a year, compared to 30 out of 10,000 not taking hormones, said Dr. JoAnn Manson, chief of preventive medicine at Brigham and Women’s Hospital. That’s a difference of only eight women in 10,000 a year – certainly significant, but depending on a patient’s risk profile, maybe not enough to justify ending a prescription.

9. Listen closely even to non-numerical claims on TV ads. This is the advice of Dr. Richard Kravitz, a professor of medicine at the University of California, Davis. An ad might say, “There is no treatment more effective than drug X,” he said. “But that may be true because nobody has looked.” An other common ad statement is, “Drug X has been proven effective for condition Y, when the real truth may be that 70% of people with the condition got better by themselves and 75% with the drug. “So the drug is effective, but not that effective,” he said.

10. The most important tip of all? If a drug or procedure sounds too good to be true, it probably is.

Ms. Foreman is a lecturer on medicine at Harvard Medical School. Her columns are available at www.myhealthsense.com.