We retrospectively evaluated 432 patients (336 black; 78%; and 96 white; 22%) incident to our peritoneal dialysis (PD; 195 patients; 45%) and hemodialysis (HD; 237 patients; 55%) programs from January 1987 to December 1997 who survived their first 90 days of dialysis therapy. Black patients comprised 70% of the PD and 84% of the HD patients (P: < 0.01). PD patients were more often men and younger than HD patients and less often had diabetes (40% versus 56% of HD patients; P: < 0.01) and cardiac disease (44% versus 58% of HD patients; P: < 0.01) than HD patients. Adjusting for baseline clinical and comorbid features, patient survival was determined by Cox regression analysis. Survival was better on PD therapy overall (relative risk [RR] for PD versus HD, 0.80; 1-, 2-, and 5-year survival rates, 90%, 77%, and 43% on PD versus 88%, 72%, and 35% on HD, respectively; P: = 0.21) and among black patients (RR for PD versus HD, 0.69; 1-, 2-, and 5-year survival rates, 92%, 80%, and 52% on PD versus 88%, 74%, and 40% on HD, respectively; P: = 0.09), but these were not statistically significant. The RR for PD versus HD was 1.08 for white patients (1-, 2-, and 5-year survival rates, 82%, 61%, and 23% for PD versus 82%, 62%, and 24% for HD; P: = 0.79). Significant predictors of mortality were race (RR for whites versus blacks, 1.51), age (RR, 1.03), cardiac disease (RR, 1.57), baseline albumin level (RR, 0.60), baseline serum creatinine level (RR, 0.91), baseline blood urea nitrogen level (RR, 1.01), and baseline weight (RR, 0.98). In conclusion, patient survival on dialysis therapy is significantly better for black patients and for patients entering dialysis with signs of adequate nutrition (increased weight and creatinine and albumin levels) and without evidence of cardiac disease. In an urban dialysis program, we find that adjusted patient survival on PD equals or is better than that on HD therapy, particularly among black patients, making PD a viable alternative to HD in our patient population.