WASHINGTON — Being overweight boosts the risk of dying from diabetes and kidney disease but not cancer or heart disease, and carrying some extra pounds appears actually to protect against a host of other causes of death, federal researchers reported Tuesday.

The controversial findings, based on a detailed analysis of decades of government data about more than 39,000 Americans, suggest that being overweight does carry risks, but the dangers may be less dire than experts thought.

“The take-home message is that the relationship between fat and mortality is more complicated than we tend to think,” said Katherine Flegal, a senior research scientist at the Centers for Disease Control and Prevention in Atlanta, who led the study. “It’s not a cookie-cutter, one-size-fits-all situation where excess weight just increases your mortality risk for any and all causes of death.”

The study, published in the Journal of the American Medical Association, was greeted with sharply mixed reactions. Some praised it for providing persuasive evidence that the dangers of fat have been overblown.

“What this tells us is the hazards have been very much exaggerated,” said Steven Blair, a professor of exercise science, epidemiology and biostatistics at the University of South Carolina. “It’s just not as big a problem as people have said.”

But others dismissed the findings as fundamentally flawed, saying an overwhelming body of evidence has documented the risks of being either overweight or obese.

“It’s just rubbish,” said Walter Willett, professor of epidemiology and nutrition at the Harvard School of Public Health. “It’s just ludicrous to say there is no increased risk of mortality from being overweight. … From a health standpoint, it’s definitely undesirable to be overweight.”

Flegal and her colleagues raised the possibility two years ago that being overweight was less risky than feared. Their analysis of data from decades of federal surveys concluded that people who were overweight — but not obese — had lower overall mortality rates than those of normal weight.

In the new research, the team sought to confirm and expand on the original findings, examining additional data from later surveys and parsing individual causes of death across a range of weights. The analysis is based on the best health statistics federal scientists collected between 1971 and 2004, including cause-of-death data from 2.3 million adults from 2004.

The most surprising finding was that being overweight but not obese was associated only with excess mortality from diabetes and kidney disease — not cancer or heart disease. Moreover, the researchers found an apparent protective effect against all other causes of death, such as tuberculosis, emphysema, pneumonia, Alzheimer’s and injuries.

An association between being overweight and nearly 16,000 deaths from diabetes and kidney disease was overshadowed by a reduction of up to 133,000 deaths from all other noncancer, nonheart disease causes. Even moderately obese people appeared less likely to die of those causes.

Although the study did not examine why being overweight might guard against dying from some diseases, Flegal said other research has suggested that extra heft might supply the body with vital reserves to draw upon to fight illness and aid recovery.

Willett and others, however, said the study was too small to fully account for factors such as smoking and pre-existing illnesses that could skew the findings. They added that being overweight not only increases the risk for dying from many major diseases but also has many other consequences.