We evaluated the association of weight and bone mass in elderly male and female subjects of the Framingham osteoporosis study, a subset of the Framingham study cohort. By examining the differences in the correlations of weight with bone mass among men and women in weight-bearing and non-weight-bearing sites and weight change since early adulthood, we attempted to understand different ways in which weight or body mass index affects bone mass. During biennial examination 20 of the Framingham cohort (1988-1989), 693 women and 439 men (mean age 76 years) had proximal femur bone mineral density assessed by dualphoton absorptiometry (DPA) and radius bone mass assessed by single-photon absorptiometry. The majority of these subjects also had spine measurements by DPA. Subjects had been weighed repeatedly over 40 years. After adjusting for other factors affecting bone density, we found that both recent weight and body mass index explained a substantial proportion of the variance in bone mineral density for all sites in women (8.9-19.8% of total variance, all p < 0.01) and for only weight-bearing sites (femur and spine) in men (2.8-6.9% of total variance, all p < 0.01). For bone mineral density at the proximal radius, weight and body mass index accounted for < 1% of variance in men (p NS). Weight change since biennial examination 1 (1948-1951) was the strongest explanatory factor for bone mineral density among women at all sites, but weight change did not affect radius bone mineral density in men. The effect of weight and of weight change on bone mineral density was in general much less in men than in women. Our results suggest that the strong effect of weight on bone mineral density is due to load on weight-bearing bones sexes. The sex difference is unexplained but may be due to adipose tissue production of estrogen in women after menopause.