Comparisons of clinical outcomes in hemodialysis (HD) and peritoneal dialysis (PD) patients have been marked by inconsistent results depending on the population studied and the methods used. In order to address this limitation of previous U.S. studies and to more specifically evaluate the higher-risk elderly population, we undertook a study of Medicare patients > or =67 years of age and assessed the comorbidity before they entered end-stage renal disease (ESRD) treatment. We then evaluated their survival outcomes at 6 month intervals in the follow-up period. In order to adequately assess the comorbidity we employed the Charlson comorbidity index and applied it to the comorbidity of the ESRD population up to 2 years before ESRD to characterize conditions from the start of ESRD treatment. We also counted inpatient hospital days in the 2 years prior to initiation of ESRD therapy as a marker of severity of disease. These two determinants of comorbidity were used to adjust the analysis along with other demographic and laboratory data. In the diabetic population, HD patients are shown to have a decreased risk of death, with the decrease ranging from 8% [relative risk (RR) (HD:PD) 0.82, 95% confidence interval (CI) 0.75-0.90] at month 6 to 54% [RR (HD:PD) 0.46, 95% CI 0.30-0.70] at month 48. In the nondiabetic population, HD patients are shown to have a 17% [RR (HD:PD) 1.17, 95% CI 1.07-1.28] increased risk of death in the first 6 months, and a decreased risk of death from months 6 to 48, a decrease ranging from 17% to 34%. Relative risks were significantly different from 1.0 at all intervals. These overall findings suggest that in the elderly population in the United States treated with PD had outcomes that were significantly worse than their HD patient counterparts, even after adjusting basic patient demographics, the comorbidity index, severity of disease with hospital days, demographics, and glomerular filtration rate (GFR) at the time of start of dialysis.