The extent to which renal allotransplantation - as compared with long-term dialysis - improves survival among patients with end-stage renal disease is controversial, because those selected for transplantation may have a lower base-line risk of death. In an attempt to distinguish the effects of patient selection from those of transplantation itself, we conducted a longitudinal study of mortality in 228,552 patients who were receiving long-term dialysis for end-stage renal disease. Of these patients, 46,164 were placed on a waiting list for transplantation, 23,275 of whom received a first cadaveric transplant between 1991 and 1997. The relative risk of death and survival were assessed with time-dependent nonproportional-hazards analysis, with adjustment for age, race, sex, cause of end-stage renal disease, geographic region, time from first treatment for end-stage renal disease to placement on the waiting list, and year of initial placement on the list.