I don't recall Rip's take exactly, but I would say (and he might even agree with me on this one) that BMI is somewhat useful for risk-stratifying health outcomes in a sedentary population. It is of limited use in people doing strength training. My view is that this is because sedentary people with high BMIs are also sarcopenic (and therefore have high BF%), which is perhaps more of what puts them at risk than the actual weight showing on the scale. A trained individual with lots of lean body mass will have a lower BF% and will be healthier than a sedentary person at the same height/weight/BMI.

Waist circumference and BF% are not the ultimate reference for healthiness by any means, but both are more useful indicators of health risks than BMI, especially in a strength training population.