We've all heard that there's an obesity epidemic in the U.S. And if you're one of the many people who has been told that you're "overweight," you've probably had at least one talk with a doctor about your Body Mass Index (BMI).

So what is BMI? Supposedly, it's an index that allows us to find the healthy weight for our height. You can even check your BMI by typing your height and weight into one of the many BMI calculators on the web. We've been told that if it's 25 or above, we'd better cut the calories and hit the gym or we're likely to pay for it with health problems later in life.

Now the risks of obesity are real. But can we rely on a single number to tell us if someone is overweight? How reliable is BMI really?

The answer is that BMI works reasonably well at what it was first designed for: looking at broad trends in the health of large groups -- to see if there's an obesity epidemic. It was never designed to assess the "healthy weight'' of an individual, so it really only sort-of works at that.

How can something work for a group of people but not for a person in that group? The answer is that as long as it works for most people (say 2/3 of the population), the errors will tend to cancel out so the trend will be valid. When BMI was first used to track obesity in groups, it was shown to work about as well as the more difficult process of calculating body fat percentage.

But the current BMI formula makes a very strange assumption. It assumes that the best way to guess how much fat you carry is to divide your weight by the square of your height (your height, multiplied by itself). This takes into account the fact that we need to get wider as we get taller, but it seems to miss the fact that we also need to get thicker, from back to front. In other words, the current BMI formula appears to assume that we're all two dimensional, as though we're cut out of cardboard.

How well does BMI work in practice? Well, if you're between about 5'5" and 5'9", it's not bad. On the other hand, an NFL quarterback in top condition might be 6'2", 220 pounds and so have a BMI of 28. His doctor might tell him he's overweight, but this person doesn't carry a lot of fat. A standard BMI chart also shows us that a person who is 6'8" is considered "overweight" at 230 pounds and "normal" at 170 pounds. That can't be right.

At the other extreme, my daughter is 4'4" and 70 pounds That's a very average height and weight for a 9-year-old girl. But pediatricians will say she's overweight because her BMI is 18. Huh?! What happened to 25? Well, BMI has to be adjusted for age in children and, the truth is my daughter is not 9 years old, she's only 6. A BMI of 18 is fine if you're 9, but it's too high for a 6-year-old (even if that 6-year-old is as tall as an average 9-year-old). How does any of that make any sense at all?

BMI is supposed to correct for height, but you can see that it doesn't unless you're close to the average adult height. For everyone who's not, I have a solution: recalculate BMIs, only this time account for all three of our dimensions. Squaring our height is appropriate for two-dimensions, but for those of us here in the 3D world, we should multiply in our height one more time (to the third power). It's not perfect, but I think it's far better than what we've got now, because this formula doesn't assume that we're cut out of cardboard.

I've created a chart (here) using the revised formula (weight/height). Since the formula is different, all of the BMI numbers have changed. Using this chart, a person with a BMI of 9 is probably underweight at any height. A person with a BMI of 10 or 11 is definitely skinny, but probably OK if they have a small frame. A BMI of 12-13 is good for a person with an average build, and a large-framed person would still be healthy with a BMI of 14. Past that, unless you're a super muscular athlete, you probably need to lose a few pounds.

Going by this chart you can see that someone who is 4'6" tall and weighs 74 pounds would have a BMI of 13. So would someone who is 5' even at 101 pounds, 5'6" at 135 pounds and 6'6" and 222 pounds. And for genuine athletes, a 6'8" NBA star who weighs about about 250 pounds would have a BMI that's between 13 and 14.

Now even this chart shouldn't be used as an absolute indicator of whether a person's weight is ideal for their height. Other factors such as a person's body habitus and how much muscle and fat they have should always be considered.

Still, BMI is being used to define an epidemic in a country. Also, insurance companies are relying more on BMIs to assess a patient's health risk and doctors are now required to asses the BMI of every patient we see. If we're going to rely on this index for so much it might help to at least bring it into the 3D world and finally start calculating it more sensibly.