NO WOMAN can be too rich or too thin, Wallis Warfield Simpson, Duchess of Windsor, is reputed to have said. The late Duchess was very rich and very thin, but no wizard in health matters.

So it vexes JoAnn Manson, researcher at Brigham and Women's Hospital and Harvard University, to see her study on the dangers of obesity reduced to brittle aphorism in the headlines. Besides, it isn't true. ``I certainly do not believe that you can't be too thin,'' she says. ``I believe you can be too thin through inadequate nutrition, excessive exercise, or smoking.''

What Dr. Manson says she and her colleagues in the Nurses Health Study meant to show was that obesity is dangerous, and overweight people have a higher risk of dying of cardiovascular disease and cancer. Study participants weighing 15 percent less than average had the lowest death rates.

And these weren't just any thin women; they were thin women who had never smoked and whose weight was stable. Longevity, it seems, favors those with slim genes over struggling dieters.

It's not news that excess fat strains the heart and raises blood pressure and the risk of diabetes, stroke and certain cancers. Obesity is a big problem in the U.S., where one-third of adults are at least 20 percent overweight.

But being too thin can also take a physical toll. Apart from deadly eating disorders like anorexia, being underweight _ say 20 percent below average _ may bring risks, too: of osteoporosis, hormonal disruption, falls, fractures and retarded healing.

The thinner a woman is, the lower her bone density, says Miriam Nelson of Tufts University. Fat fuels estrogen production, and its loss can make bones porous and brittle. Also, heavier bodies create denser bones; lighter bodies, lighter bones.

WHEN COUPLED with overexercise, undernourishment can further disrupt monthly hormonal cycles, leading to missed menstrual periods or infertility. It's a syndrome often seen in ballerinas and marathon runners who drive their body fat to subnormal levels.

With frailty, older people _ both men and women _ face increased risks of falling and fractures, adds Dr. Nelson. Weight charts lose meaning for people over 70, where low fat sometimes means low muscle mass _ and lost strength and coordination.

``To say you can't be too thin is simplistic _ and wrong,'' she says.

Remember the skeletal socialites that novelist Tom Wolfe dubbed ``social X-rays?'' asks Dr. Nelson. ``Those are the women at risk for falls and fractures.''

Dianne Budd of Women's Health Access at the University of California at San Francisco worries that the Harvard study may wrongly tempt some women into radical diets or fasting _ which won't help and might harm. She contends the Harvard study narrowed its sights on ``a magical subset of women'' who were genetically programmed to be healthily and stably thin.

``They're about 10 percent of the population,'' says Dr. Budd. ``And none of those are my patients!'' To get to be 15 percent below average weight _ and stay there _ most people would need to diet rigidly, a course she calls ineffective and, at worst, unsafe.

Restricting food intake lowers the body's metabolic rate _ the speed at which you burn calories _ making some feel tired and cold. It also alters measures of satiety _ that feeling of satisfaction that helps curb the appetite naturally. All this may cause a weight rebound when the diet ends.

``If normal-sized people limit calories, depression goes up, and there's a decrement in critical thinking skills. People wonder why they're so spacey. And a certain number of people do progress to eating disorders,'' says Dr. Budd. Focus instead on heart-healthy diets and regular exercise, she urges.

Undernourishment also depletes nutritional reserves needed to fight infection. Stanford University dietician Jo Ann Hattner says being 20 percent underweight ``is where I'd begin to consider that someone is at nutritional risk.'' An ultra-thin person may have slower recuperation.

POUNDS DON'T TELL the whole health story, contends C. Wayne Calloway of George Washington University, a proponent of more lenient weight charts. ``Weight alone isn't a good predictor of mortality until you get to extremes,'' he says. ``Some people are more muscular than others.'' He contends mid-size people who are energetic, well-nourished and regular exercisers shouldn't feel they're at risk just because they aren't in the lowest 10th percentile for weight.

``If you're genetically lean and otherwise healthy and active, there's nothing wrong. If you're lean because you're smoking, drinking or engaging in severe caloric restriction, then there are some major problems,'' Dr. Calloway says.

What many health experts agree on is that muscle tone, energy, exercise and endurance should get equal time with the number on the scale in discussions of health. It's how we function that counts, not just how much we weigh.

How important is that 10- to 20-pound weight gain?

Dr. Manson says it's a ``warning signal'' that warrants a change to a healthier lifestyle. But Dr. Budd argues someone who is 5 feet 5 inches and weighs 140 pounds won't necessarily live longer by dieting down to 120. By the study's definitions, they're no longer in that golden group of stably slender.

No one denies the overwhelming evidence against obesity, but neither is it healthful to enshrine an ideal of unattainable slenderness.