So what changed between this study and the two that reached the opposite conclusion? The methodology, of course. The first two relied on body mass index (BMI), a rather crude means of calculating obesity based simply on a person's height and weight. As a result, people with above-average muscle mass are sometimes inappropriately classified as overweight.

Furthermore, the previous research did not distinguish between smokers and nonsmokers. Those with cigarette habits tend to weigh less. Aune thus excluded smokers entirely from his sample. He also excluded those with long-term health conditions.

Among all nonsmokers, the mortality risk was observed at BMIs between 23 and 24, the study found. Among "healthy nonsmokers," the magic numbers were between 22 and 23. Finally, among those whose medical records spanned more than two decades, those with BMIs between 20 and 22 were the least likely to die.

As the obesity rate in the U.S. and elsewhere has steadily risen over the past three decades, medical authorities have generally agreed that our expanding waistlines represent a public health crisis.

However, there is less agreement on how to effectively combat obesity, with many experts now claiming that years of emphasis on reducing fat intake, smaller portion sizes, and physical fitness has been flawed. Instead, they argue, the culprit is sugar and refined carbohydrates.