It is unclear whether increased muscle mass or body fat confer the survival advantage in hemodialysis patients with high body-mass index (BMI). Twenty-four-hour urinary creatinine (UCr) excretion was used as a measure of muscle mass. The outcomes of hemodialysis patients with high BMI and normal or high muscle mass (inferred low body fat) and high BMI and low muscle mass (inferred high body fat) were studied to study the effects of body composition on outcomes. In 70,028 patients who initiated hemodialysis in the United States from January 1995 to December 1999 with measured creatinine clearances reported in the Medical Evidence form, all-cause and cardiovascular mortality were examined in Cox and parametric survival models. When compared with normal BMI (18.5 to 24.9 kg/m(2)) group, patients with high BMI (> or = 25 kg/m(2)) had lower hazard of death (hazard ratio [HR], 0.85; P < 0.001). However, when compared with normal BMI patients with UCr >25th percentile (0.55 g/d), high BMI patients with UCr >0.55 g/d had lower hazard of all-cause (HR, 0.85; P < 0.001) and cardiovascular death (HR, 0.89; P < 0.001), and high BMI patients with UCr < or =0.55 g/d had higher hazard of all-cause death (HR, 1.14; P<0.001) and cardiovascular death (HR, 1.19; P <0.001). Both BMI and body composition are strong predictors of death. The protective effect conferred by high BMI is limited to those patients with normal or high muscle mass. High BMI patients with inferred high body fat have increased and not decreased mortality.