Chronic allograft nephropathy (CAN) in protocol biopsies is associated with graft loss while the association between subclinical rejection (SCR) and outcome has yielded contradictory results. We analyze the predictive value of SCR and/or CAN in protocol biopsies on death-censored graft survival. Since 1988, a protocol biopsy was done during the first 6 months in stable grafts with serum creatinine <300 micromol/L and proteinuria <1 g/day. Biopsies were evaluated according to Banff criteria. Borderline changes and acute rejection were grouped as SCR. CAN was defined as presence of interstitial fibrosis and tubular atrophy. Mean follow-up was 91 +/- 46 months. Sufficient tissue was obtained in 435 transplants. Biopsies were classified as normal (n = 186), SCR (n = 74), CAN (n = 110) and SCR with CAN (n = 65). Presence of SCR with CAN was associated with old donors, percentage of panel reactive antibodies and presence of acute rejection before protocol biopsy. Cox regression analysis showed that SCR with CAN (relative risk [RR]: 1.86, 95% confidence interval [CI]: 1.11-3.12; p = 0.02) and hepatitis C virus (RR: 2.27, 95% CI: 1.38-3.75; p = 0.01) were independent predictors of graft survival. In protocol biopsies, the detrimental effect of interstitial fibrosis/tubular atrophy on long-term graft survival is modulated by SCR.