Diagnosis of allograft dysfunction relies on the assessment of arterial lesions. This study was designed to evaluate the prognostic significance of common specific vascular lesions in acute allograft rejection. Renal allograft biopsies (n = 111) with acute cellular rejection were scored for endarteritis, mononuclear cell adherence to endothelial cells, endothelial activation, fibrinoid necrosis, foam cells, and intimal fibrosis. These vascular lesions and other classic histologic features were correlated with outcome. Rejection with endarteritis (found in 54% of biopsies) was less responsive to steroid treatment than rejection without endarteritis, as judged by recovery of creatinine in 3 wk (P = 0.03). Larger numbers of sampled arteries improved the predictive accuracy. Sticking of mononuclear cells to endothelial cells also correlated with steroid resistance (P < 0.05). Rejection with or without endarteritis responded to OKT3/antithymocyte globulin treatment equally well (61% versus 65%, respectively). Rejection with fibrinoid arterial necrosis (4% of biopsies) did not respond to either steroids or antibodies (0%). One-year graft failure was 21% without endarteritis, 28% with endarteritis, and 100% with fibrinoid necrosis. Activated endothelial cells and interstitial hemorrhage were associated with endarteritis and graft failure (all P < 0.05). None of the other scored features had any statistically significant correlation with outcome. Thus, specific arterial lesions (endarteritis, fibrinoid necrosis, activated endothelial cells, mononuclear cell margination) and interstitial hemorrhage, but not the extent of the interstitial infiltrate or tubulitis, are correlated with response to antirejection therapy and/or 1-yr clinical outcome. Grading systems for therapeutic trials and clinical management should emphasize scoring of specific vascular lesions.