We analyzed the five-year survival of 4661 patients with end-stage renal disease whose first dialysis occurred in 1977. The analysis incorporated characteristics of patients and dialysis institutions. Results showed that mortality was positively correlated with patient age, initial conditions leading to renal failure, being male and white, open staffing, and the number of staff physicians. In addition, lower death rates were observed for patients treated in larger dialysis units and units that had been long-term reusers of dialyzers. Patients treated in for-profit and not-for-profit units appeared to have the same mortality, although patients treated in freestanding units had lower mortality. The direction of causation was not always clear in these results.