Mortality among end-stage renal disease patients in the United States remains unacceptably high despite progress in the management of renal replacement therapy. Consequently, there are few reports of long-term survivors on dialysis. We have analyzed characteristics of long-term (10 to 15 years, N = 40) and very long-term (15 to 30 years, N = 18) survivors on hemodialysis and long-term survivors (more than 10 years, N = 28) on peritoneal dialysis and compared them with "average survivors" (< 5 years, N = 65 for hemodialysis and N = 101 for peritoneal dialysis). Among hemodialysis patients, long- and very long-term survival was associated with younger age, nondiabetic status, black race, and male gender (P < 0.05 for all variables). Enrollment creatinine was higher among long- and very long-term survivors, whereas albumin and hematocrit increased significantly during the period of observation among long- and very long-term survivors compared with average survivors. Enrollment age, nondiabetic status, and albumin level predicted prolonged survival even after adjustments for confounding variables. Among peritoneal dialysis patients, younger age and nondiabetic status predicted prolonged survival. Black race was associated with improved survival, but the association was not statistically significant. Enrollment levels of albumin and creatinine were significantly higher among long-term survivors and the cholesterol increased during the period of observation in long-term survivors. Thus, demographic and biochemical indices reflecting nutritional status can predict prolonged survival in hemodialysis and peritoneal dialysis. Patient survival for periods of up to 30 years is possible on renal replacement therapy. Analyses of these outlier patients may offer clues to prolonged survival.