OBJECTIVE:Cross-sectional studies show that active individuals are leaner than their sedentary counterparts. The purpose of this paper is to determine the extent that this is due to self-selection bias, specifically to initially leaner men and women choosing to exercise longer and more intensely. METHODS:Walking quantity (weekly distance) and intensity (speed) were compared to current body mass index (BMI) (BMI(current)) and BMI at the start of walking (BMI(starting)) in 20,353 women and 5174 men who on average had walked regularly for exercise for 7.2 and 10.6 years, respectively. RESULTS:The relationships of both BMI(current) and BMI(starting) to both distance and intensity were nonlinear (convex). On average, BMI(starting) explained >70% of the association between BMI(current) and intensity, and 40 and 17% of the associations between BMI(current) and distance in women and men, respectively. Although the declines in BMI(current) with distance and intensity were greater among fatter individuals than leaner individuals, the portions attributable to BMI(starting) remained relatively constant regardless of fatness. Thus, self-selection bias accounted for most of the decline in BMI with walking intensity and smaller, albeit significant, proportions of the decline with distance. CONCLUSION:Although walking intensity and walking distance are both strongly associated with greater leanness, intensity is less likely to be causally related to leanness than is distance, and the latter should be encouraged for reversing or preventing weight gain.