Just considering weight, this gives nice reproducible readings without "cheating" and just redisplaying its last reading if it suspects you're the same recent person, which my old Salter ones did. Might vary by 0.1kg in successive readings, but that means it's being honest. As a scrawny marathon runner, I found impedance measurements of body fat gave a huge overestimate in the past. This is pretty good though -- reckons about 12%, against a DEXA result I had 3 years ago of 9%, so that's pretty good. It may well have gone up a touch (more like 9-10% from skinfold though). This morning I measured my weight, BF% and water%, then did a 28 mile run and measured again. Weight went down from 58 to 54.8kg (very plausible), BF% down from 12.2 to 10.9% and water% up from 62.5 to 63.2%. That "increase" in water sounds suspicious but actually that's a water mass decrease of 1.6kg, and a fat decrease of 1.1kg. No way would I have burnt 1.1kg of fat but quite likely 1.1kg of fat + carbohydrate, so it's kind of doing OK in this one-off test. Bad points: user interface is completely non-intuitive. Fine once you know what you're doing. But no button is labelled ON -- to turn it on you either have to press USER and then SET to do the works, or OFF/WEIGHT just for weight. Intuitive huh?