Risk factors for bone loss among the elderly are largely unknown. The objective of the study was to examine longitudinal bone loss in the hip in one-hundred and sixty-two 75-year-old women. Bone mineral density (BMD, g/cm(2)) was measured with dual X-ray absorptiometry (DXA) at baseline and after 4 years. The relationship between changes in BMD during follow-up and the following factors; baseline BMD, baseline weight, weight change, baseline lean and fat body mass (measured with DXA), serum values of biochemical markers and hormones, nutritional and lifestyle factors according to a questionnaire was assessed. The annual mean (SD) change in femoral neck BMD was -0.31% (1.38) in total trochanter -0.35% (1.15) and total hip -0.34% (1.10) and did not differ significantly between measurement sites. Bisphosphonate users had a 2.9%, 1.7% and 1.9% mean adjusted increase in femoral neck, total trochanter and total hip BMD respectively, different from none-users (p<0.05). Subjects with more than three weekly physical activity sessions had less femoral neck bone loss than less active women (p<0.05). The proportion of the variance in BMD changes explained by multivariate models (R(2)) was 12-13%. Women gaining weight had less loss of BMD than those losing weight in the trochanter and the total hip (p<0.001), and in the femoral neck (p=0.055). Elderly women should be advised to maintain their body weight and participate in physical activity. Despite the large number of variables examined in this study, bone loss occurring with increased age is not thoroughly explained.