Both endogenous and exogenous glucocorticoid excess are well establish as causes of osteoporosis, however the reversibility of bone mass following the restauration of normal steroid levels is not well documented. In this longitudinal study, we mesured bone mineral density (expressed as Z-score) by dual-photon and X-ray absorptiometry of the lumbar spine (L2-L4) in 9 patients cured of Cushing's syndrome who were followed for the next 48 months (16-76). Initial Z-scores ranged from -2 to -1 standard-deviation (SD) in 6 patients consistent with osteopenia, and were below -3 SD in 2 patients consistent with osteoporosis. One patient developed lumbar spine fractures. There is no relationship between the severity of the Cushing's syndrome (assessed by the urinary free cortisol) and initial bone reduction (inital Z-score), nor between length of Cushing's symptoms and initial bone reduction. Our data show a marked variation (+74 +/- 9%) in bone mass in patients successfully treated for Cushing's syndrome. Seven patients completely recovered from steroid-induced osteoporosis, one patient partially recovered but remained osteopenic. One post-menopausal women presented several lumbar spine fractures despite successfull treatment of Cushing's syndrome. This longitudinal study confirms that if steroid-induced bone loss may improve substantially by cure of steroid excess even without other treatment, osteoporosis may worsen particularly in post-menopausal women. These results are important to take into account to properly manage patients with steroid-induced osteoporosis.