Well, I've been challenged on the 95% of diets fail statistic, even though I had seen it cited on literally hundreds of sites, all referencing a National Institures of Health 1992 Consensus Conference. After much digging, I finally found the text of the Conference Paper here.

Having read the source material. Although the paper in question does state "most people who achieve weight loss with any of these programs regain weight," I find no mention of an actual failure rate; therefore, I apologise for disseminating the same misinformation everyone else has.

That said, the paper in question makes all sorts of conclusions that even it admits haven't been scientifically proven. It takes as given that any relationship between weight and disease is causal, with absolutely no scientific data to back that up. It mentions the BMI and states that the government and the scientific community "have suggested slightly different desirable ranges of BMI." It also states that the "prevalence of overweight has increased during the last two decades" without any raw data on which to assess that statement. Are they are talking raw numbers? Percentages? By what standards? How much of that increase is due to the decrease in the officially sanctioned "healthy" BMI? Without knowing that, it's anybody's guess as to accuracy of the conclusions the panel drew.

The study did conclude although dieting does help with certain diseases like diabetes mellitus and hypertension in th elong run, "[l]ong-term health effects are much less clear."

It also concluded that that more scientific study is needed and stated that
"[o]besity in humans has a substantial genetic basis," calling for research into the genetic and environmental causes of obesity. It also stated that since "weight loss [is] associated with increased mortality," better scientific understanding of weight loss and its affects is needed.

"[W]eight loss [is] associated with increased mortality" - i.e. dieting can kill you.