Computed tomography (CT) scanning was used to assess the relationship of glucose tolerance to fat distribution in men. Three cross sections [chest (including upper arms), abdomen, and thigh] were scanned in 41 men randomly selected from the Normative Aging Study, a longitudinal study of aging. Greater amounts of fat in the upper body and greater ratios of upper-body fat to lower-body fat were significantly correlated with higher 2-h serum glucose levels after adjustment for age and body mass index. In particular, intra-abdominal fat, a feature uniquely measured by CT, was a significant correlate of 2-h glucose. Largely parallel results were obtained when we compared a sample of male diabetic subjects (N = 8) with the male normal subjects from our random sample. This investigation demonstrates that body fat distribution, adjusted for overall degree of obesity, is a significant correlate of glucose tolerance even in a sample unselected for extremes of physique.