The numbers are in on obesity, and the results might shock you: slightly overweight people have a lower mortality rate than those with a healthy body mass index.
Being chubby has its benefits
It is not just the health and number of fat people that draw concern these days. The science underpinning the very notion of fatness is looking pretty sick as well.
Researchers are starting to concede that a significant portion of "overweight" people may be healthier than they think. They also suspect a similar portion of those who think they are at an ideal weight may be in much worse shape than they know.
Doubts about what constitutes an unhealthy weight have been circulating for decades. It hardly helps that much of the early research on it was so shambolic. The link between body fat and an early death was popularised by United States life insurance companies in the 1960s, who noticed that people who were a few pounds heavier than average seemed to die earlier. Suddenly millions of Americans found themselves classified as "overweight", and their desperation to avoid an early grave sparked the rise of fad diets, exercise regimes and slimming pills.
Those who tried them soon found that the whole issue of weight, diet and exercise was far more complex than the "experts" were claiming. Dieters discovered that even if they could shed a lot of weight, they could not keep it off once they stopped dieting - and often ended up weighing even more after a few months.
Exercise proved similarly disappointing, while diet pills turned out to have nasty, even lethal, side-effects.
By the early 1980s, a more fundamental issue emerged. Some scientists began to question whether all the slimming, sweating and pill-popping was even necessary.
It became clear that the early work linking weight with life expectancy was dogged by "confounders" such as smoking, age and gender which blurred the real link between weight and longevity.
Using more sophisticated analysis, researchers made a bizarre discovery - the people classified as overweight seemed to have the lowest risk of premature death. This cast serious doubt on the method used to assess risk from their weight, the so-called body mass index, or BMI.
This required a calculation in which the weight of a person in kilograms is divided by the square of their height in metres. Those whose resulting BMI exceeded 25 were overweight, while those giving a figureof 30 or more were deemed clinically obese.
Plotting the risk of premature death against BMIs gave a U-shaped curve - which made some sense, as being incredibly thin and having a low BMI is no healthier than being extremely fat with a huge BMI.
What made less sense, however, was the discovery that the lowest point of the U-shaped curve - that is, the healthiest BMI - was not always at 25. Studies suggested that as we get older, we may benefit from being a bit paunchy, with BMIs of more than 25 being healthier for those in middle age or older than lower BMIs.
The most compelling evidence for this emerged in January this year in research published by a team led by Katherine Flegal of the US National Centre for Health Statistics in Maryland.
Pulling together the results of almost 100 studies covering almost 3 million people, the team concluded that those classified as "overweight" - that is, BMIs between 25 and 30 - have mortality rates 6 per cent lower than those in the supposedly healthy BMI range of about 19 and 25.
These findings have been met with vociferous criticism, with some scientists arguing that the study's statistical methods were flawed because they lumped data for many different ages and genders in to one ovrall data set.
There is, however, a growing consensus that at the very least our reliance on BMI to gauge healthy weight needs rethinking.
According to Dr Rexford Ahima and Dr Mitchell Lazar at the University of Pennsylvania in Philadelphia, the problem lies in the fact that BMI fails to reflect new research showing that not just weight but how it is distributed, and how our fat content compares to muscle content, are important as well.
In a review published in the journal Science last month, the researchers point out that "visceral fat" hidden within the abdomen has been linked to potentially lethal diabetes and cardiovascular diseases.
In contrast, more visible "belly fat" can serve as a safe store for toxins - and an energy reserve during times of illness. Yet the BMI does not distinguish between them.
As a result, say the researchers, about 10 per cent of people in the US deemed obese may actually be relatively healthy, their weight coming from the less lethal, but more visible, form of fat.
And the reverse may also hold, with a similar proportion of "normal weight" people actually being at higher risk than they think, because of their invisible store of more dangerous visceral fat.
It is clear that the science linking weight to health is becoming increasingly complex, with a host of factors such as fat distribution, gender and even ethnic origin playing a key role.
For example, people of Asian origin tend to carry more fat around their middle, making them at greater risk than Europeans.
But while researchers wrestle with this complexity, the rest of us are left without any simple rule of thumb by which to gauge our weight.
Fortunately, researchers are working on alternatives to BMI. The simplest is waist measurement: men of any height should be concerned if this exceeds 94 centimetres, and the limit is 80cm for women.
A more sophisticated measure is the body shape indicator (BSI), which factors in waist size as well as height and weight. It also gives a simpler link with mortality, without the strange U-shaped curves perplexing scientists. Just stay below a certain BSI figure, and you are likely to be fine.
What scientists have yet to find, however, is a way of making that easy. Nobel Prizes await the first to succeed.
Robert Matthews is a visiting reader in science at Aston University, Birmingham, England