ne of the central tenets of modern medicine is that the earlier your doctor can catch a disease, the better. This has proved a brilliant strategy for several conditions. Detecting high blood pressure, for example, and treating it can cut a patient's risk of stroke by a third and the chances of heart disease by 20 percent. Identifying patients with the first signs of diabetes and controlling their blood sugar can help significantly reduce the chances of cardiovascular disease, kidney and neurological damage and blindness. By spotting an abdominal aneurysm, doctors can monitor the progress of the defect and repair it before it ruptures. The notion that disease could be prevented at all was shaped by the sanitary movement of the 19th century, when doctors and antipoverty activists set about improving the living conditions of the poor and reduced the appalling rates of infectious disease. Today, the simplest and most effective prevention medicine stops disease before it starts through vaccinations, for instance, or encouraging people to exercise more, smoke less, eat fruits and vegetables and drink in moderation -- all of which can cut the risk of cancer, cardiovascular disease, diabetes and stroke.

But while this sort of prevention clearly works, a troubling trend has taken root when it comes to doctors actively treating patients with drugs or procedures to prevent disease. Dr. Eric Schneider, an assistant professor of medicine at the Harvard School of Public Health and Harvard Medical School, refers to it as ''prevention creep.'' It is the general faith in the power of medical intervention to thwart not only certain discrete diseases, like hypertension, but also most chronic or potentially life-threatening conditions. The corollary to this idea is that intervening works best when the disease is detected as early as possible, preferably before symptoms appear. Doctors and their patients, Schneider says, have come to believe implicitly in the notion that catching any disease early and treating it -- often aggressively -- can make you live longer.

Schneider was the lead investigator on a study that looked at angioplasty, a common procedure that is often viewed as a preventive measure. Angioplasty is meant to treat the physical symptoms of heart disease -- shortness of breath and angina pectoralis, or chest pain -- and can be a lifesaver when administered within the first few hours of a heart attack. During the procedure, an operation that can be done with local anesthesia, a doctor threads a special wire fitted with a tiny deflated balloon, no wider than a pencil lead, through the abdominal artery up into the coronary arteries, which supply blood to the heart. On a cine-angiogram, a kind of X-ray movie, you can literally see a clot blocking the blood flow to the heart. That clot is squashed against the wall of the artery when the balloon is blown up. Once the balloon is deflated, the clot breaks up and blood begins flowing into the heart muscle. Alternatively, cardiologists can install a stent, or steel mesh tube, which is expanded inside the artery and left in place.

In 2000, a balloon or a stent saved the lives of many Americans who were given an angioplasty while in the midst of a heart attack. But most of the million angioplasties performed were done on people who were not in imminent danger of dying. Schneider's Harvard study suggests that more than two-thirds of these so-called elective angioplasties (not done to save the patient's life, but to prevent future symptoms) were performed on patients who, by criteria accepted by cardiologists themselves, may have stood to gain little or nothing from the procedure.

When Schneider's group looked at 828 angioplasties done on Medicare beneficiaries, they found that only about a third of the patients could have expected to benefit from the procedure. Another 14 percent of the angioplasties were completely inappropriate, either because the patients had no physical symptoms or the narrowing in their arteries was so slight it could not possibly be causing any problems or their condition was so severe that they needed cardiac bypass operations. For the remaining half of the patients in the study, it was impossible to decide if they stood to benefit.

Why would cardiologists want to subject patients to an elective treatment they might not need? Part of the answer is that huge numbers of Americans -- about two million a year -- are being given angiograms, the imaging test that shows narrowings, or stenosis, in the coronary arteries. Once cardiologists see a narrowing, it is very difficult not to want to do something about it. Dr. Eric Topol, chairman of the department of cardiovascular medicine at the Cleveland Clinic, calls it ''the occulostenotic reflex.'' ''You see a narrowing, and you think, We need to do a stent,'' he says. ''That is a problem, because so many narrowings are innocent -- they don't need a stent; they aren't tied to symptoms in the patient. But you get on the train, and you can't get off once it's moving.'' Too many cardiologists, Topol says, do not do enough to treat the chronic condition, the underlying problems like high cholesterol that are causing the narrowing. They need to encourage their patients to adopt a healthier lifestyle, for instance, and to prescribe anti-cholesterol medication and aspirin. ''There is a big force to do angioplasty procedures,'' he says. ''It can be done quickly; it is quite safe; and cardiologists find it relatively lucrative -- and that's not helping matters.'' But also driving many elective angioplasties is the unspoken hope that the operation will be a magic bullet, a quick and safe procedure that will prevent a heart attack sometime in the future.

Unfortunately, at this point there is no evidence to support that idea. It is wishful thinking that angioplasty prevents future heart attacks, explains Dr. Robert Lager, a cardiologist at the Washington Hospital Center in the District of Columbia. ''But patients want to believe it. You can see the look in their eyes when you tell them that this is a chronic disease. And it's easy for us as doctors to believe it too. But angio was designed to treat symptoms.''

Angioplasty is a template for the manner in which many procedures came into widespread use in American medicine's laudable but often flawed effort to prevent disease. First, doctors discover that a procedure saves lives or effectively treats an acute condition. Next they begin applying it to patients whose disease is not so advanced in the hope that they will keep the condition from worsening. And then they begin looking for earlier and earlier signs of the disease in patients who feel perfectly healthy. Prevention has become the doctor's watchword, the mantra that drives the use -- and often misuse -- of procedures and diagnostic tests in every branch of medicine from cardiology to psychiatry. Its pursuit has blurred the boundaries of medical futility and helped fuel the conviction shared by doctors and patients that it is better to err on the side of doing more rather than less.

elief in this sort of interventionist preventive medicine has ascended in lock step with the proliferation of diagnostic tests, which allow doctors to detect signs of disease in people who have few or no symptoms. Twenty years ago, the seemingly healthy middle-aged man who went for a physical would have his blood pressure checked, along with blood and urine tests. If he complained of chest pains, he might get on a treadmill for a stress test. A woman would get a Pap smear and possibly a mammogram.

Today, the number of tests has exploded. ''We no longer just treat sick patients, but instead we look at well patients to find evidence of disease,'' says Dr. Stephen Baker, chairman of the radiology department at the New Jersey Medical School. A routine checkup for the average middle-aged American will likely include a P.S.A. test for prostate cancer for men, the DEXA scan for osteoporosis, a Pap smear and high-resolution mammography for women and a cholesterol test and colonoscopy for everybody. Radiologists have recently begun setting up Doppler ultrasound booths in local malls, where you can have your carotid arteries checked for blockages. Then there's the CA-125 test for ovarian cancer, calcium screening for cardiovascular disease and spiral CT scan for lung cancer. And that's only a partial list.

A few diagnostic tests -- for instance, the Pap smear, the colonoscopy and the DEXA scan for osteoporosis -- have proved to be effective tools for defensive medicine, allowing doctors to catch disease and treat it before symptoms appear. But many other diagnostic tests, which have their places if a patient has symptoms, are given routinely to apparently healthy people in the name of prevention. They are having the perverse effect of benefiting only a small minority of patients while exposing the majority to invasive, often risky, treatment they don't necessarily need.

This is true for mammography and the prostate specific antigen, or P.S.A., test -- two screening tests that most Americans unvaryingly believe will help thwart an untimely death. On the surface, the logic seems obvious. Prostate cancer is the second most common cause of cancer death among men, after lung cancer, and breast cancer ranks second for women. Both tests can catch tumors when they are tiny. High-resolution mammography can now detect breast cancers not much bigger than two letters printed on this page, while the P.S.A. test catches prostate tumors on average 11 years before a digital rectal exam. But if early diagnosis really worked, then the mortality rate, or deaths per 100,000 in the population, should go down as more and more people are screened. Some cancer epidemiologists and doctors now argue that there is little evidence that either screening test has lowered the death rate significantly for its respective cancer.

That is because tumors come in many types, only some of which will never become aggressive enough to threaten the patient's life. Many tumors will sit around for years, causing few if any symptoms, and the person will die of something else before the cancer becomes a problem. On the flip side, many dangerous tumors cannot be cured with current treatments, no matter how early they are caught. Unfortunately, radiologists and pathologists can't tell with much certainty how aggressive a particular tumor is going to be. ''Imaging has improved so much, we can find things that we really don't know enough about,'' says Dr. William Black, a radiologist at Dartmouth Medical School. In the face of this uncertainty, doctors say they must err on the side of caution and treat practically every tiny tumor as if it were potentially deadly in the hope of curing at least a few.

But that means that widespread screening for prostate and breast cancer has resulted in huge numbers of patients suffering the side effects of unnecessary medicine. A study published last October in BMJ (formerly the British Medical Journal) compared P.S.A. testing rates in the region around Seattle and in Connecticut between 1987 and 1997. Men in Seattle were 5.39 times as likely to get a P.S.A. test as men in Connecticut and 5 times as likely to undergo surgical removal of the prostate. Yet there was no significant difference in the death rate from prostate cancer between the two regions. Prostate removal is a delicate surgical procedure that renders many patients impotent, incontinent or both. In a study of patients at Harvard University hospitals, where you might expect the best outcomes, the majority of men were impotent a year after their surgery.

It is less clear that mammograms are not helpful. But there is still little evidence that they are. Seven different clinical trials conducted since the 1960's, involving well over 100,000 women in the United States, Canada and Europe, have tried to determine the benefits of mammography. The most recent analyses of those trials have found critical flaws in their design, leaving an increasing number of cancer experts to conclude that there is no clear evidence to suggest that mammography is saving lives.

The same cycle is coming into play for detecting kidney cancer. More and more tumors are being spotted on the kidneys by the full-body CT scan, a new technology that is being heavily marketed by large academic medical centers and hundreds of freestanding clinics around the country. But as with breast or prostate tumors, many of these tumors may turn out to be, on closer inspection, simply benign lumps that appeared to be cancer at first. Unfortunately, closer inspection usually means a surgical procedure. Baker cites a study that indicates that 85 percent of kidney spots that show up on a scan will turn out to be false-positives. He also explains that a 40-to-60-year-old man who undergoes a full-body scan has a 1 in 20,000 chance of having malignant kidney cancer at the time. When doctors start looking for disease in healthy people, Baker says, ''the odds are in favor of making mischief rather than a cure.''

Take the DEXA, or dual-energy X-ray absorptiometry scan, which is one of the screening tests that when used properly can help prevent disease. Doctors routinely use the DEXA on women 65 and older to diagnose osteoporosis, or thinning bones, one of several risk factors for fractures in the elderly. When the DEXA scan first came into use a decade ago, a panel of international experts convened to decide the cutoff point for diagnosing osteoporosis, drawing the line at bone density that was a certain level below that of the average 25-year-old woman. The panel's standard has helped doctors to identify women at highest risk for fractures and treat them with drugs like estrogen and Fosamax, which slow bone loss.

Doctors get into trouble when they try to stem osteoporosis in younger and healthier women. At the same time that the expert panel set the standards for osteoporosis, it also invented a new term, ''osteopenia,'' for women who had suffered a little bone loss but not enough to put them in the category of osteoporosis. The panel hoped that women would be screened once soon after menopause and that a diagnosis of osteopenia would serve as a warning sign for them to stop smoking and step up their efforts to get enough calcium and exercise, primary prevention that can slow the loss of bone. Instead, a diagnosis of osteopenia has led many doctors to prescribe Fosamax and other drugs immediately rather than waiting for osteoporosis.

''Doctors and their patients don't even know what osteopenia means,'' says Susan Love, a breast-cancer surgeon and frequent critic of many postmenopausal treatments. There is no evidence to suggest that earlier treatment will be any more effective in arresting fractures than waiting for true osteoporosis. In fact, some animal studies now under way are suggesting quite the opposite, that long-term use of the drugs may cause bones to become more brittle, Love says, ''and we are prescribing these drugs to women who might or might not suffer a fracture 15 or 20 years down the road.''

But the most troubling possibility involves procedures performed in the name of prevention, whose potential risks far outweigh any possible benefits. The carotid endarterectomy, for example, is a kind of Roto-Rooter surgery for the arteries that supply blood to the brain. In use since the 1950's, carotid endarterectomy is performed on about 150,000 Americans a year to avert strokes that are caused when a blood clot forms in one of the two carotid arteries in the neck and blocks the flow of blood when it reaches a narrower arterial branch in the brain. Carotid endarterectomy is the most common vascular surgery performed in the United States, and according to the National Stroke Association's Web site, the procedure reduces the risk of stroke ''by as much as 55 percent.''

But this figure is accurate only for the patient whose carotid arteries are at least 70 percent blocked and who has already experienced either a full-blown stroke or a ministroke, a temporary loss of blood flow to part of the brain. Epidemiologists estimate that 3 out of 10 people who fall into this category will suffer a stroke within five years. Undergoing carotid endarterectomy reduces their risk to 1 in 10. But most of the endarterectomies in this country aren't performed on these high-risk patients, but rather on people whose chances of stroke are significantly lower. In January, Dr. Peter Rothwell, a British neurologist, published results from a study of more than 6,000 patients pooled from several large European and American clinical trials. The study confirmed that the operation really does help avert strokes in people at the highest risk. But for patients who had only moderately blocked arteries the surgery's benefit is offset by the 6 to 7 percent risk of stroke or death posed by the procedure itself. As for patients with only mild narrowings, the procedure increased the possibility of stroke.

As the Doppler screening test becomes increasingly omnipresent, more and more people are being tested for blockages. Those tests often lead to procedures. And two-thirds of carotid endarterectomies performed in this country are on people who have no symptoms. That means that last year about 100,000 Americans underwent a procedure that not only had little chance of benefiting them; it also caused some of them serious harm. At an average cost of $15,000 per endarterectomy, that is $1.5 billion the health care system could have put to better use.

Though some doctors are doing carotid endarterectomies on patients who will not benefit, how should any doctor make a decision about the patients who are in that gray area where there is no evidence that a test or treatment will actually save their lives, but also no evidence that it won't? And how do you keep patients from demanding those interventions in order to be better safe than sorry? Some doctors are now testing a new method of care called ''shared decision making,'' which is showing that the choices patients make are connected to what they know. Rather than trying to determine what the best course of action is -- whether to perform an angioplasty, say, or a P.S.A. test -- a doctor gives the patient a videotape to watch at home that lays out in clear and simple terms what is known about the test or procedure being considered, and what the potential dangers are. Patients, it turns out, are often more wary of the risks of medical intervention than their doctors are. In a recent study conducted at the Dartmouth-Hitchcock Medical Center in Hanover, N.H., researchers compared two groups of people with severe back pain caused by a ruptured disc. One group simply talked to their surgeons about having a back operation. The other group took home a video that described the possible risks and benefits. The group that saw the video was 30 percent less likely to choose surgery compared with the other patients.

Shannon Brownlee is a senior fellow at the New America Foundation.