1. In 1966, the half-life (1-year posttransplantation) for first cadaver-donor kidney grafts reported to the UCLA Registry was around 7.5 years (1). Between 1985 and 1990, this half-life was 7.8 +/- 0.02 years. Since 1966, the corresponding 1-year graft survival rates increased by over 30 percentage points. Clearly, improvement in early graft survival has had little bearing on long-term graft outcome. 2. From a stratified multivariate analysis of 40,582 primary cadaveric renal transplants recorded in the UCLA Transplant Registry from 3 consecutive eras (1975-79, 1980-84, 1985-90), the following long-term graft survival trends in covariates have emerged: a) a constant strong negative effect associated with higher numbers of HLA-A,B mismatches, younger and older recipients, diabetics, and longer cold ischemia times in each era; b) an increased beneficial effect on female recipients; c) an increased detrimental effect on Black recipients, despite short-term gains; and d) a positive effect of CsA usage (only in the most recent era). 3. From a multivariate analysis of 15,027 primary cadaveric renal transplants reported to the UNOS Kidney Registry between 1987 and 1990, we analyzed the effects of transplantation factors on survival during 3 consecutive posttransplantation risk periods: 0-1 month; 1-3 months; and beyond 3 months. Few pretransplant factors affected risk of failure within 1-month posttransplantation. However, a good predischarge clinical course (as indicated by CsA usage, no required dialysis during the first postoperative week, and no rejection episodes) was associated with an immediate improvement in graft survival. The effects of most UNOS transplantation factors during the second risk period were comparable to the short-term coefficients estimated from the UCLA file; and the effects of the UNOS factors on "beyond 3-month" risk were comparable to the UCLA long-term coefficients. Conclusively, the dominant pretransplant factor on long-term risk was HLA-A,B tissue matching.