High physical activity level has been associated with high bone mass and low fracture risk and is therefore recommended to reduce fractures in old age. The aim of this study was to estimate the effect of potentially modifiable variables, such as physical activity, muscle strength, muscle mass and weight, on bone mass in elderly women. The influence of isometric thigh muscle strength, self-estimated activity level, body composition and weight on bone mineral density (dual energy X-ray absorptiometry; DXA) in total body, hip and spine was investigated. Subjects were 1004 women, all 75 years old, taking part in the Malmö Osteoporosis Prospective Risk Assessment (OPRA) study. Physical activity and muscle strength accounted for 1-6% of the variability in bone mass, whereas weight, and its closely associated variables lean mass and fat mass, to a much greater extent explained the bone mass variability. We found current body weight to be the variable with the most substantial influence on the total variability in bone mass (15-32% depending on skeletal site) in a forward stepwise regression model. Our findings suggest that in elderly women, the major fracture-preventive effect of physical activity is unlikely to be mediated through increased bone mass. Retaining or even increasing body weight is likely to be beneficial to the skeleton, but an excess body weight increase may have negative effects on health. Nevertheless, training in elderly women may have advantages by improving balance, co-ordination and mobility and therefore decreasing the risk of fractures.