According to a study by the American College of Sports Medicine, it's the Minneapolis-St.Paul metropolitan area, followed by Washington, D.C.:
Link via Marginal Revolution | Photo (unrelated) by Flickr user Jason Lengstorf used under Creative Commons license
Their winning rank reflects the cities' relatively low (and rapidly-diminishing) smoking rate, their above-average percentage of regular exercisers, moderate-to-low rates of obesity, asthma, diabetes, and other chronic concerns, and rising share of farmers' markets (indicative of a trend towards healthier dining). Boston takes the bronze, with Portland, Oregon fourth and Denver in fifth place. At the opposite end of the spectrum, Oklahoma City ranks as America's least fit metro, followed by Louisville, Memphis, Birmingham, and Detroit.
Link via Marginal Revolution | Photo (unrelated) by Flickr user Jason Lengstorf used under Creative Commons license
There are a number of metabolic variants that affect the way adipose tissue is stored in the body. For example; Peroxisome Proliferator-Activated Receptor Gamme (formerly Delta) converts stored adiposity into muscle tissue and in some people this process is accelerated making them fairly lean regardless of fat consumption, whereas in others it functions poorly and such people are generally obese according to the BMI and culture in general. Even if these people go on an all vegetable and fruit diet they will still be obese. Their problem is not what or how much they eat, but the way their PPAR-Gamma receptors respond to PPAR-Gamma. Diet, exercise, pump iron, they will still be fat and society at large will still ridicule them for being fat.
Now, we also know that genetics share a correlation with geography in many species. Members of a species in one region will have a specific variant which only barely bleeds over into the next region.
I'm sure genetics is a big factor in fitness, but it's not the only factor. And you would still need to take fairly large gene samples from a variety of populations and locations, to separate the genetic factors from the environmental ones.
Short of that, factors like behavior, and more easily measurable traits, like obesity, are going to be the most useful data.
And while BMI is pretty bunk as an absolute obesity measure, especially for people with muscular builds, it's still a useful variable. Lower BMI still correlates to fitness in large groups of people, since muscular builds are a relative minority in most representative samples.
Also, a person's fitness is not set in stone at birth. And until we engage in genetic engineering on ourselves, it would behoove us to focus on the behavioral factors at play. There's an obesity epidemic on, you know... and unless you think that there has been a genetic mutation in the last 100 years, I think it's our major improvement in prosperity, and the huge behavioral shifts in the last few generations that are responsible.
Finally, most of the areas compared are not ones of genetic isolation. I can testify anecdotally, that since moving from San Diego to Portland, my lifestyle has changed markedly in response to the different environment. The food tends to be fresher and healthier here, and people are much more active and fit. I lost weight and got markedly fitter since I've moved here.