A historical prospective national sample of 1,725 diabetic and 2,411 non-diabetic Medicare end-stage renal disease (ESRD) patients incident from 1986 to 1987 was analyzed for the mortality of patients selected to receive continuous ambulatory peritoneal dialysis (CAPD) or hemodialysis (HD) with adjustment for patient characteristics, including the presence of comorbid conditions at onset of ESRD. Cox proportional hazards analyses were used to compare the mortality of CAPD and HD patients. Patients were followed from 30 days following onset of ESRD until two to four years post-onset. No statistically significant difference in relative mortality risk (RR) was found among non-diabetic patients selected for CAPD compared to HD (RR = 0.84 for CAPD versus HD, P = 0.25), while evidence of higher adjusted mortality for CAPD compared to HD was found among diabetic patients (RR = 1.26, P = 0.03). Mortality analyses adjusted for pre-treatment risk factors suggest that CAPD and HD provide incident non-diabetic ESRD patients with similar expected survival outcomes. Evidence that increased mortality was associated with CAPD among diabetic patients, particularly among elderly patients, suggests the need for further controlled studies of mortality among CAPD patients with diabetes.