There is a little debate over mortality and body mass index in humans. The overwhelming majority of epidemiological evidence coupled with animal studies and what is known of the underlying biochemistry shows that being fat is bad for you on a sliding scale: the more fat, the worse off you are. The occasional study turns up to claim the opposite, which is the way things go in science. It doesn't matter how overwhelming the evidence or solid the theories, it is still possible for professional teams to generate contrary data in good faith, by accident or simply via the whims of statistical chance. So last year a study was published suggesting that the moderately overweight do best in old age, having the lowest mortality rates. This was widely debated at the time and, I think, largely successfully dismantled and shown to be a poor result.
Nonetheless, the wheels of scientific publication move slowly, and that published data is referenced in the interesting paper on calorie restriction linked below, which was probably written before last year's debate wound to a close. Regardless, the general thrust of the paper remains worth reading, which is how to reconcile various short and long term data obtained from the study of calorie restriction, some of which appears at first glance to be contradictory. The practice of calorie restriction greatly improves health and extends healthy life spans in animal studies. In human studies it is shown to greatly improve short term measures of health. As to the few primate studies, there is a great deal of debate over what exactly the results of twenty years of data are in fact showing us, and whether that research was carried out in a way that allows all of the useful conclusions we'd like to see. Under the hood, there are many theories and much evidence on how calorie restriction alters biochemistry to slow aging, but no unified theory as of yet. It is probably very complex. Calorie restriction is a jigsaw with many missing pieces at every level, from the biochemistry all the way up to drawing lines between the results in different species.
Still, calorie restriction is well supported as a beneficial thing to be doing as a human. Eat less and benefit. The big question is whether this is "just" the best thing you can presently do for your long-term health while paradoxically having little effect on your life expectancy, or whether it can add additional years. The present scientific consensus is that it probably won't add more than five to seven years to life, but is just as good as regular moderate exercise in many ways and considerably better in others.
As this paper shows, there are those who don't really look too far beyond simple measures over populations, such as BMI and mortality. I think they are doomed to see only paradox, and whatever resolutions exist are to be found in the lower level details. At some point there has to be a good resolution between detailed observations that show large benefits from calorie restriction, a mountain of statistical population data telling us that being overweight is bad, and a much lesser collection of statistical population data telling us that being overweight isn't bad. Meanwhile, I'll still with the compelling evidence for not eating so much and keeping the fat tissue to a minimum.
A recent, very large meta-analysis has shaken the epidemiological community by showing that the lowest inflection point for the BMI-mortality curve (its nadir) lays in the overweight range. Discussion is ongoing among epidemiologists on this topic, some time referred to as the "obesity-mortality paradox." There are several confounding factors, in fact, to consider: for example, smokers tend to weigh less but have higher mortality; some chronic diseases may induce weight loss; the frail elderly with higher risk of death may also experience weight loss, etc.
An important question that gerontologists and epidemiologists should try to answer together is the following: "people who voluntarily choose a CR regimen and are already within a normal BMI range, let us say its upper half, are increasing their longevity or their mortality?" Indeed, when glancing on reports about members of the Calorie Restriction Society, or CRONies (Calorie Restriction with Optimal Nutrition) as they call themselves, we should consider their BMI. In one of the longest studies available, for example, where subject were monitored for a period of 6 years, 28 weight-stable CRONies had an average BMI of 19.7, and they were compared with 28 age-matched subjects on a typical western diet who had an average BMI of 25.6 and served as the control group. These two groups, for example, had BMI values that quite precisely spanned the normal BMI range. If these two groups of persons will keep their body weight constant for the future, what could we predict regarding their longevity? Using as guidance studies like the [the meta-analysis noted above], we should conclude that the control group should experience a lower mortality. Instead, using as guidance the generally accepted idea that CR extend laboratory animals life span together with the few available prospective studies where persons who were leaner in youth or in midlife resulted longer lived, we should conclude that the CRONies will actually experience decrease mortality and extended longevity.
Which somewhat skips over the point that the calorie restriction practitioners in that study had a longevity-associated measurement of biochemistry that was much better than the control group members. To me this is the interesting paradox regarding calorie restriction: why is it that the shifts in biochemistry are so similar between humans and mice, and yet only the mice experience a sizable extension of life as a result?