In order to evaluate the intersample variations for bone histomorphometric parameters in various metabolic bone diseases, either for a group or for one single patient, two complete contiguous transiliac bone biopsies were taken in 55 subjects. The diagnoses were osteoporosis (OP), renal osteodystrophy (ROD), osteomalacia (OM), primary hyperparathyroidism (HPT), metastatic bone disease, fluorosis, thyrotoxic bone, and "normal" bone. The following histomorphometric parameters were measured: trabecular bone volume (TBV), trabecular osteoid surfaces (TOS) and volume (TOV), trabecular resorption surfaces (TRS), and calcification rate (CR). The thickness index of osteoid seams (TIOS) was calculated. The measurements were performed with both manual and computerized methods which give similar results according to our previous study. The differences between parameters values measured on both cores were expressed by the difference in percent of the mean and by the intrapair coefficient of variation. Moreover, for each parameter, the confidence interval for one subject was calculated from the residual mean square of a two-way analysis of variance. For each parameter, the intersample variation differs according to the diagnosis. Confidence interval (risk = 5%) for one single subject reaches 29% for TBV in OP; 16% for TOS; 26% for TOV and TIOS in OM; 25% for TRS in ROD, and 69% in HPT, but is much lower for groups of 10 and 20 patients. These variations must be taken into account when successive biopsies are performed in one individual or in groups of patients to follow the course of a disease or evaluate the effects of a therapy.