It is a useful guideline for the general public. If it says you're overweight and are sporting a beer belly, you likely aren't an edge case.

Better measurements exist, but for most collecting the data needed is complex (and the impedance measuring ones aren't reliable). If you have a suggested alternative I'd welcome reading about it.

It is only used for historical reasons. It's really bad science, made popular by Ancel Keys whose reputation for bad science is impeccable.

Waist to Height ratio is AFAIK the best "simple but data supported" measure: https://qz.com/1002707/bmi-calculators-arent-accurate-but-ou...

He's right - the BMI formula penalizes very muscular people and tall people. The height issue comes from the square term in the formula, and obviously it cant distinguish muscle from fat.

Nick Trefethen from Oxford:

https://people.maths.ox.ac.uk/trefethen/bmi.html

It's not, if you understand what BMI actually means.

BMI is an aggregate metric that has been misapplied to individuals.

Due to the averaging effect, it's a pretty decent predictor for groups, but due to variation within groups, it's not going to predict that well for all individuals (except for individuals who are close to the average of the group).

Someone with 12% bodyfat is not going to be in any danger of confusing themselves for being overweight. For average people it is a good rule of thumb.

Obviously there are better tools, but none of them are so easy to measure.

How are you measuring bodyfat? I'm interested in tracking it, but seems tricky to do accurately.

DXA scan is pretty much the only way to do it accurately. Calipers, "smart" digital scales, and comparing yourself to diagrams online are all very poor methods of checking bodyfat and should be considered to have somewhere around a +-10% error range.

I found a guy in my city doing DXA scans for $80, I usually go get one done every six months or so.

Edit: Nevermind, stupid question. Thanks for the corrections!

BMI isn't a percent, but based on your height/muscle mass you can have an overweight BMI with very low body fat.

BMI is not a percentage. The unit for BMI is kg * m^-2.

Ancestor post was referring to either percentage of total body mass as fat, or percentage of lean mass. Both are dimensionless, as kg/kg cancels out.

As such, if the 12% was % of lean mass, that would be 10.7% of total body weight. If it was 12% of total body mass (more likely), that would be 13.6% of lean mass. Either way, that's classified as "very lean".

BMI = kg/m^2. (Weight divided by height). It doesn’t account for the composition of the person whatsoever.

The 12 was body fat percentage, not BMI.

Yep, Dwayne "The Rock" Johnson is also obese based on BMI numbers alone.

A world-class athlete with a near superhuman physique may be the perfect definition of an edge case in this situation.

I’m neither of those and fall into the same “edge case”.

12% body fat with enough muscle to almost be considered obese by BMI standards is quite a few standard deviations outside of the mean. Anyone with that kind of body is well aware of that BMI shortcoming. For most of the population, it’s still a pretty good rule of thumb.

There's a very good chance you're not testing your body fat correctly. I've gotten several friends, including bodybuilders and various athletes to get DXA scans done... pretty much every one underestimated their BF% by 5-10%. Including the ones who swore up and down by their personal trainer's caliper measurements, the BF testing machine (the electric handles one) at their local supplement store, etc.

I'm 14% BF and a BMI of 20... looks to be right on point.