Background. Chronic rejection (CR) remains idiopathic, difficult to prospectively identify, and once detected, unresponsive to increased immunosuppression. We hypothesized that clinically stable human renal allografts have ongoing evidence of injury and immune activity, and that this correlates with the worsening of allograft function characteristic of CR. Methods. The allografts of 40 stable renal allograft recipients were biopsied 2-3 years after transplantation. Biopsies were processed for histology and RNA extraction. RNA was evaluated by semi-quantitative RT-polymerase chain reaction for CD3γ mRNA (a marker of T cell receptor turnover), and mRNA from cytokine genes previously shown to be transcribed during acute rejection: tumor necrosis factor-α, interferon-γ interleukin- (IL) 1β, IL-2, IL-4, IL-6, and IL-8. Clinical parameters including urine protein and glomerular filtration rate were measured the day of biopsy. Findings were then compared with clinical outcome to establish associations between subclinical inflammation and graft dysfunction. Allograft function was measured again 2 years after biopsy and correlated with findings at the time of biopsy. Results. Cytokine transcripts and histological evidence of injury were detected in more than two-thirds of stable grafts. The degree of the lymphocytic infiltrate correlated with the degree of proteinuria (P=0.034) and histological fibrosis (P=0.005). Similarly, the degree of intragraft CD3γ transcription correlated with increasing proteinuria (P=0.043). IL-6 and IL-8 transcripts were also correlated with evidence of graft injury. After 2 years, those biopsies originally found to have evidence of fibrosis, tubular atrophy, or CD3γ transcription had worsening graft function as determined by creatinine and glomerular filtration rate. Conclusions. These data demonstrate that significant injury and immune activity can be detected in patients who are stable on clinical grounds. Undetected subclinical graft injury may be a cause of chronic allograft rejection.