Based on a log-linear model for analyzing multiple effects, center variation in the UCLA Transplant Registry is the most influential factor in early renal graft survival. Among 68 active centers accounting for 70% of all transplants between 1985 and 1989, the 1-year graft survival (adjusted for 15 covariates) varied from 60 to 90%. The long-term graft survival, as measured by the half-lives of the first cadaver donor transplants at these centers, based on a minimum of 4 years of follow-up data, varied from 3 to 26 years. Although there was some correlation (r = 0.24) of half-life with the 1-year graft survival rate at each center, in many instances, excellent centers had poor half-lives and average centers had excellent half-lives. From a Cox regression analysis on data from 6752 first cadaver-donor renal transplants surviving beyond 1 year (submitted to the UNOS data base), white female recipients of 0 HLA-AB-mismatched kidneys not requiring dialysis in the first week posttransplant had projected half-lives of 14.1 years, whereas black male recipients of 4 HLA-AB-mismatched kidneys requiring therapeutic dialysis had half-lives of 2.3 years. However, adjustment for these factors could only remove 17% of the total variability in the centers' half-lives. Consequently, any algorithm intended to judge a center's renal transplant capabilities should include, even after adjustment for covariates, some function of both 1-year graft survival and half-life beyond 1 year.