OBJECTIVE: Waist and hip circumferences have been shown to have independent and opposite associations with glucose levels. Waist circumference is positively associated with glucose levels, whereas hip circumference is negatively associated. It is unclear which tissues are involved in the pathophysiological mechanism causing these associations. The main goal was to determine which tissue in the trunk and legs, fat or lean tissue, is associated with measures of glucose metabolism. RESEARCH DESIGN AND METHODS: In 623 participants of the third examination of the Hoorn Study, whole-body dual-energy X-ray absorptiometry was performed to determine fat and lean soft-tissue mass in the trunk and legs. Fasting and 2-h postload glucose levels after 75-g oral glucose tolerance test (OGTT) were determined. After exclusion of known diabetic patients, cross-sectional analyses were performed in 275 men aged 60-87 years (140 with normal glucose metabolism, 92 with impaired glucose metabolism; and 43 with diabetes) and in 281 women (148 with normal glucose metabolism, 90 with impaired glucose metabolism, and 43 with diabetes). RESULTS: Greater trunk fat mass was associated with higher glucose levels after adjustment for age, trunk lean mass, leg lean mass, and leg fat mass. Standardized beta (95% CI) in men were 0.44 (0.25-0.64) for fasting and 0.41 (0.22-0.60) for postload glucose. For women, these values were 0.49 (0.35-0.63) and 0.47 (0.33-0.61), respectively. In contrast, in the same regression models, a larger leg fat mass was associated with lower glucose levels. Standardized beta in men were -0.24 (-0.43 to -0.05) and -0.12 (-0.31 to 0.07) and in women -0.24 (-0.37 to -0.10) and -0.27 (-0.40 to -0.13) for fasting and postload glucose, respectively. In these models, larger leg lean mass was also associated with lower glucose levels but was only statistically significant in men. CONCLUSIONS: If trunk fat is taken into account, accumulation of fat in the legs seems to be protective against a disturbed glucose metabolism, ...