Okay, with the proviso that I haven't actually read the study in question (only the abstract), it would appear that the results are based on simple correlations between BMI and cause of death.

All of the conditions that overweight allegedly "protects" you from are conditions that themselves cause weight loss! So, it's not that overweight protects you from tuberculosis, it's that anybody who does of tuberculosis usually loses a ton of weight before dying. Any serious, uncontrolled infection generally causes weight loss. Ditto for cancer. Ditto for emphysema (people burn lots of calories due to the metabolic costs of laboured breathing.)

Severe Alzheimers and Parkinsons also cause weight loss because the afflicted are unable to feed and care for themselves.

Unfortunately, the growing trend in medical journals is to publish any study that boasts large enough numbers without any consideration of the scientific merit behind them.

This thing should have been published in the journal "Duh".

Quite right. Also, as haiku_rd notes, smokers tend to be thin. Smoking lowers appetite, and smokers use niciotine rather than food or some other addictive substance to regulate anxiety. Thus being moderately overweight is not shown to be protective by this study, but may simply be a correlate of not being a smoker, and being a smoker does lead to higher mortality rates. A difference in mortality rates for thin smokers may have produced the observed results either a) with there actually being some protective value due to a few excess pounds or b) in spite of there also being a real positive correlation between being moderately overweight and higher risk of health problems.

I agree also. That is why I think we should not be so quick to blame those who are obese for a correlational relationship. Perhaps their diseases such as diabetes cause weight gain or increases in appetite. Why do we continue to waste our time and money on these correlational studies and start doing some real research?

Totally agree. And why do scientific reporters at major media outlets, and here on Slate a columnist whose only job is to discuss scientific news, not point out these obvious competing explanations while repeatedly giving space and attention to these meaningless correlation studies? Boggles the mind ...

and, as another reader pointed out, BMI is not the same as a fat percentage calibration. a person with a substantial amount of muscle will look like a fatso despite having very little actual body fat. for instance, my BMI makes me look like i'm a few pounds shy of being overweight but am easily 5 dress sizes smaller than my friends who weigh less but never hit the gym.

I find it funny that in another thread there were people ranting about the obese smokers who were huge health risks and then in this thread, everyone is asserting that smokers are thin all of a sudden. I don't think smoking really has to do with your weight all that much. I've seen fat smokers and thin smokers. Just wanted to point that out...

Funny how being overweight "protects" you mainly from diseases that occur later in life . . . if you die of a heart attack at 40, you won't get Alzheimers . . . some "protection!"