This study was designed to evaluate the pathologic criteria used for acute renal allograft rejection that were developed by a panel of renal pathologists participating in the Cooperative Clinical Trials in Transplantation, a National Institutes of Health-supported, multicenter research group. The panel defined three categories of acute rejection. (1) Type I: mononuclear infiltrate in > or =5% of cortex, a total of at least three tubules with tubulitis in 10 consecutive high-power fields from the most severely affected areas, and at least two of the three following features: edema, activated lymphocytes, or tubular injury. (2) Type II: arterial, or arteriolar, endothelialitis with or without the preceding features. (3) Type III: arterial fibrinoid necrosis or transmural inflammation with or without thrombosis, parenchymal necrosis, or hemorrhage. Using these criteria, and without any knowledge of the clinical course or original diagnosis, a rotating panel of three pathologists agreed with the original study pathologist's diagnosis of the presence or absence of rejection in 259 of the 286 biopsies (91%) used for this analysis (kappa = 0.80). The sensitivity to establish the diagnosis of rejection was 91% for a single core and 99% for two cores. To validate the diagnostic criteria, the thresholds for number of tubules with tubulitis and the percent infiltrate were varied, and the pathologic diagnosis was compared with the clinical course. The greatest agreement occurred with a threshold of > or =1 tubule with tubulitis and > or =5% cortex with interstitial infiltrate (91%). Clinically severe rejection episodes were correlated with the type of rejection (type I, odds ratio [OR] 6.2; type II, OR 37.9). Type II rejection was more likely to be clinically severe than type I (OR 6.1). Analysis of other individual pathologic features revealed a correlation with clinical severity for endothelialitis (OR 13.2), interstitial hemorrhage (OR 13.2), and the presence of glomerulitis (OR 3.7) (all P < 0.05). The extent of tubulitis or of the interstitial infiltrate did not correlate with severity (P > 0.05). It is concluded that these criteria are simple, reproducible, and clinically relevant. These data should lead to further refinement of the diagnostic systems for renal allograft rejection.