Background. As short-term transplant results improve, it has become difficult to use patient or graft survival or acute rejection as clinical trial endpoints, except in large, multicenter studies. Despite better outcomes, graft failure continues over time. Methods. We studied 6- and 12-month creatinine (Cr) level and change in creatinine (ΔCr) level (3-12 months, 6-12 months) as predictors of graft survival for 1,389 primary kidney transplants (minimum graft survival 1 year). Determining the prognostic value of Cr level (6 or 12 months), the subgroups were as follows: less than 1, 1 to 1.4, 1.5 to 1.9, 2.0 to 2.4, 2.5 to 2.9, and greater than or equal to 3 mg/dL. For ΔCr level, the subgroups were as follows: less than 0, 0, 0.01 to 0.2, and greater than 0.2. Subgroup actuarial graft survival was determined. Cox regression analyses were performed with forward, stepwise selection. Results. After 12-month Cr level entered the model, no other variable was significant. Repeating this with continuous variables, 12-month Cr level was again the best predictor. Five-year graft survival for 12-month Cr level less than 1 (n=38) was 95%; for 1.0 to 1.4 (n=454), 87%; for 1.5 to 1.9 (n=463), 86%; for 2.0 to 2.4 (n=166), 78%; for 2.5 to 2.9 (n=54), 60%; for greater than or equal to 3 (n=45), 41%. A major breakpoint for outcome is 1-year Cr level=2.0. A power analysis was performed for the combined endpoint of graft loss and 1-year Cr level greater than 2, reached by 30% of patients. To avoid missing a reduction to 20% (actual decrease 33%) (alpha=0.05; power=0.8), 313 patients would be required per group. For a reduction to 15% (actual decrease 50%), 133 patients would be required. Conclusions. Twelve-month Cr level is an accurate surrogate for long-term outcome. The use of a combined endpoint (graft loss and 12-month Cr level) allows trials to be performed without exorbitant numbers.