Because the comparison of survival in patients with renal failure treated by dialysis and transplantation may be biased by pretreatment prognostic differences in the patients who receive these two therapies, we quantified the pretreatment prognosis of all 430 dialysis and transplant patients who began therapy for end-stage renal disease at two hospitals from 1970 to 1980. Five pretreatment factors had a statistically significant adverse effect on survival: age, duration of diabetes, left ventricular failure, myocardial infarction, and other serious comorbid illness. Dialysis patients had a worse pretreatment prognosis than transplant patients did. When we controlled for these pretreatment differences, the actuarial 5-year patient survivals were 80% for dialysis (D), 79% for cadaver transplantation (CT), and 91% for living donor transplantation (LDT), (P = 0.9 for CT vs. D, and P = 0.05 for LDT vs. D). This similarity in survival with dialysis and cadaver transplantation was quite different from the results obtained when pretreatment prognosis was not controlled; the uncontrolled 5-year patient survivals were 43% for D, 77% for CT, and 89% for LDT (P less than 0.001 for CT vs. D, and P less than 0.001 for LDT vs. D). Our data suggest that the major factor determining differences in survival with dialysis and renal transplantation is not the relative efficacy of the two treatments but the pretreatment prognostic status of the patients chosen to receive them.