The excellent present outcomes of renal transplantation pose major challenges for the construction of future studies of interventions to reduce late renal allograft loss. Using current data on expected renal allograft outcomes, we performed power calculations to estimate the required size of trials of primary and secondary interventions in chronic renal allograft rejection. A primary prevention trial would require recruitment of 1500 patients over three years and is probably not feasible. A secondary intervention trial would require entry of only 126 patients and is still feasible. The difficulty of performing trials using graft failure as the primary study outcome has encouraged interest in use of surrogate endpoints such as acute rejection. Using the UNOS renal transplant data base, we examined changes over time in the frequency of acute rejection within the first year and the risk of subsequent graft loss. Changes in acute rejection rates were neither parallel to nor predictive of changes in the rate of late graft loss, providing no support for the use of acute rejection as a surrogate for late graft loss. Future continued improvement in renal transplant outcomes may require changes in traditional scientific and administrative methods for evaluating the impacts of proposed interventions.