Osteoporosis that develops during immobilization is a severe condition that confers increased risk of fractures with their burden of mortality and disability. The aim of this study was to investigate the determinants of immobilization osteoporosis. As a model of this condition we studied hemiplegic subjects, measuring bone mineral density in the paralyzed lower limb as compared with the non-paralyzed one. In spite of the limits related to the loss of nervous stimulation, this model offers the advantage of a proper control for the complex genetic and environmental cofactors involved. We examined 48 hemiplegic subjects (31 men, 17 women in menopause) admitted consecutively over a 9-month period. Mean length immobilization was 10.9 months for men (range 1-48 months) and 7.8 months for women (range 1-40 months). The average time since menopause was 14.9 years (range 1.7-23.9 years). For each subject the following were performed: questionnaire, medical examination, anthropometric measurements, evaluation of the scores for spasticity and for lower limb motor capacity in order to account for the different degrees of disability among patients. Bone mineral density was measured using dual-energy X-ray absorptiometry (DXA) at both femoral necks. For each patient we defined a percentage difference in bone loss between the paralyzed and non-paralyzed limb. Regression coefficient were calculated by multiple logistic regression. There was significant bone loss in the paralyzed limb in both sexes, accounting for up to 6.3% in women. Multiple regression analysis showed that the degree of bone loss depends significantly and directly on the length of immobilization, even when controlling for age and sex in the regression model (R = 0.193, p = 0.034). However, when time since menopause was included in the regression model, with length of immobility as a covariate, it was the only significant determinant of bone loss (R = 0.312, p = 0.039). No additional factors were observed among men. No differences were shown with regard to anthropometric measurements or functional scores. Length of immobilization accounts only for a small fraction of bone loss, which does not exceed 5% of the total variance. Our data show that postmenopausal women should be considered at highest risk for osteoporosis in cases of immobility and that different factors, other than length of immobility, might come into play in determining bone loss in this condition.