I'd like to emphasize the "might not": The impression I get from all these studies (I've seen them pop up pretty regularly for as long as I've been paying attention, nearly two decades) is that the border is actually wrong, that going from "normal" to "overweight" should be increased slightly from 25 to 27 or so. We're probably getting mixed signals in the "overweight" classification because "okay" and "bad" levels are being combined.