To determine the value of the history and examination in predicting a temporal artery biopsy result, 81 temporal artery biopsies between 1974-1980 were studied. Each biopsy was scored blindly and independently by 2 pathologists and correlated with clinical data and the indications for biopsy at the time of the biopsy. Thirteen biopsies had giant cells and were considered positive, 8 had mononuclear inflammatory cells in the wall (7 cases) or healed arteritis (1 case) and were considered probable, and 60 were negative. In patients with an elevated sedimentation rate, the presence of a sign (tenderness or abnormal temporal artery), or a symptom (visual change or headache) predicted 18 of 21 (86%) positive-probable biopsies, but misclassified 32 of 60 negative biopsies. Attempts to derive a clinical decision rule with improved specificity resulted in unacceptably low sensitivity (missing 5 of 21 positive-probable biopsies). A minimum 5-year followup showed that those patients treated with steroids had a 6-fold increase in fractures and a 4-fold increase in cataracts.