A cross-sectional study was performed in a group of dialysis patients and control subjects to identify the determinants of serum levels of leptin, the protein product of the obese (ob) gene. Twenty-eight patients on dialysis (19 patients on hemodialysis [HD] and nine patients on peritoneal dialysis [PD]) and 41 healthy control subjects were studied. For each subject, blood was drawn for measurement of serum leptin levels and body composition was analyzed by dual-energy x-ray absorptiometry. Regression analyses were performed to determine the predictors of leptin levels, the independent contribution of HD and PD, and the relationship between leptin levels and markers of malnutrition and protein intake in the patients on dialysis. As expected, percentage of body fat was strongly correlated with leptin levels in the group as a whole and in each subgroup when analyzed separately. However, the slope of the relationship was significantly greater for dialysis patients than for control subjects (P < 0.05). Multivariate regression analysis showed that patients on HD and PD had higher leptin levels than control subjects even after adjustment for age, gender, and percentage of body fat. Univariate analyses were performed to assess the relationship between leptin levels and markers of nutritional status such as albumin, blood urea nitrogen, protein catabolic rate (PCR), transferrin, cholesterol, and lean body mass per height. There was a significant negative correlation between leptin levels and serum albumin (r = -0.598, P < 0.001) and between leptin and PCR (r = -0.433, P < 0.05) in the patients on dialysis. It is concluded that leptin levels adjusted for percentage of body fat are increased in dialysis patients compared with control subjects, particularly in those on PD. In addition, increased leptin levels are associated with low serum albumin levels and PCR in dialysis patients.