This cross-sectional twin study examined the influence of constitutional, lifestyle, and genetic factors on bone mineral density (BMD) in elderly women. BMD, at the lumbar spine, femoral neck, Ward's triangle, total hip, and total forearm, total body bone mineral content (BMC), and lean mass and fat mass were measured using dual energy X-ray absorptiometry in 69 volunteer female twin pairs (37 monozygotic [MZ], 32 dizygotic [DZ]) aged 60–89 years. Height and weight were measured. Medical history and lifetime tobacco and alcohol use were determined by questionnaire. In terms of within-pair differences, lean mass was independently associated with BMD at all sites. In contrast, fat mass was not associated with BMD at any site once allowance had been made for lean mass. Lifetime tobacco use was independently associated with BMD at the lumbar spine, total hip, and forearm. Total body BMC was independently predicted by lean mass, fat mass, tobacco use, and alcohol consumption. Age and the above independently predictive body composition and lifestyle factors accounted for 20–33% of variation in BMD. After allowing for these covariates, MZ and DZ correlations were consistent with about 75% of residual variation in BMD at the nonforearm sites being determined by genetic factors. For total body BMC, the covariates explained 75% of total variation, and genetic factors 76% of the residual variation. Therefore, at the proximal femur and lumbar spine, after taking into account the relation of BMD with lean mass and smoking, genetic factors appear to play a substantial role in explaining variation in BMD in elderly women.