It is still unclear whether dialysis modality, i.e., continuous ambulatory peritoneal dialysis (CAPD) versus hemodialysis (HD) specifically affects bone mineral density (BMD). To answer this question, 34 patients on HD and 25 on CAPD were matched for age, sex, height, and body weight with 125 normal subjects. BMD was measured using dual-energy X-ray absorptiometry (DXA; Hologic QDR 1000/W) at the lumbar spine (trabecular bone), the femoral neck (mixed cortical and trabecular bone), the distal tibial diaphysis (cortical bone), and the epiphysis (trabecular bone) in all subjects. No significant difference for blood hemoglobin, albumin, total and ionized calcium, intact parathyroid hormone (PTH) or phosphorus concentrations, as well as for alkaline phosphatase activity, failed renal allograft, prior steroid therapy, prior parathyroidectomy, duration of uremia, or of dialysis was found between patients on HD and those on CAPD. However, the residual daily urine volume and renal function at the time of the absorptiometry were higher in CAPD than in HD patients (p < 0.05) as well as the mean dialysate calcium concentration during dialysis, the blood bicarbonate concentration, and the residual renal function at the initiation of dialysis (p < 0.01, p < 0.05, and p < 0.005, respectively). In contrast, the total dose of calcium carbonate was lower in CAPD than in HD patients (p < 0.01). Results of BMD were expressed as Z scores (the number of standard deviations from the appropriate mean of BMD of 623 healthy subjects adjusted for age and sex). At the lumbar spine, no significant difference with respect to BMD was observed between the three groups. At the femoral neck and tibial epiphysis, HD patients had lower BMD (p < 0.001) than normal controls, whereas no difference was observed between HD and CAPD patients. At tibial diaphysis, patients on HD had lower BMD (p < 0.001) than patients on CAPD and than normal controls, with the values being similar in patients on CAPD and in normal controls. The results remained identical after exact matching of HD (n = 25) and CAPD (n = 25) patients for dialysis duration (1.9 +/- 0.3 and 1.7 +/- 0.3 years, respectively). Multiple regression analysis revealed significant negative correlations between Z scores at the lumbar spine (p < 0.05), femoral neck (p < 0.02), tibial diaphysis (p < 0.005), and tibial epiphysis (p < 0.05) on the one hand and plasma alkaline phosphatase activity on the other. The Z score at tibial diaphysis was also correlated with residual renal function at the initiation of dialysis (p < 0.05). In conclusion, this study provides evidence for the preservation of cortical bone with CAPD as opposed to HD. The higher residual renal function observed in the former treatment modality might account, at least in part, for this finding.