Seventy renal allograft biopsies were done in 31 patients, routinely at 1, 2, and 3 months posttransplant, and as clinically indicated, using an automated biopsy "gun." The histological diagnosis was made according to the Banff schema, which emphasizes tubu-litis and vascular inflammation over mononuclear cell infiltration. Fifty-three biopsies satisfied histological inclusion criteria. Twenty-nine biopsies were obtained from stable patients, defined as those in whom serum creatinine had changed <10% in 2 weeks, and in whom immunosuppression (cyclosporine, azathioprine, and prednisone) had not been increased in that interval. Of these biopsies, 30% (9/29) showed rejection, which could not have been predicted from pretransplant (HLA mismatch, panel-reactive antibody titer) or posttransplant (cyclosporine and serum interleukin 2 receptor levels) variables. The significance of these early subclinical rejection episodes is unknown, and their effects on long-term graft histology and function are being examined in a controlled study.