We have reviewed the clinical records and histology of 135 patients who underwent temporal artery biopsy between 1973 and 1978. Biopsies were classified histologically as giant-cell arteritis (17%), atypical arteritis (6%), healed arteritis (2%), arteriosclerosis (67%), atherosclerosis (5%), or normal (3%). Most of the histological diagnoses made at the time of biopsy were confirmed but eight cases which had originally been reported as atypical or healed arteritis were classified as arteriosclerosis when reviewed. All 33 patients with histological evidence of arteritis were accepted as clinical cases of temporal arteritis (31) or polymyalgia rheumatica (2) and treated with steroids. A further 24 patients had negative biopsies (arteriosclerosis or atherosclerosis) but were considered on clinical grounds to have cranial arteritis. They too were treated and made a full recovery. In 43 cases, all of whom had negative biopsies, a final diagnosis was reached which was thought to account for the clinical symptoms (e.g., cerebrovascular accident, rheumatoid disease, migraine, etc.). As less than 60% of patients with clinical temporal arteritis had positive biopsies, we suggest that this procedure could be omitted and replaced by a trial of steroid therapy. Biopsy should be reserved for patients with a strong medical contraindication to steroid therapy, or who fail to respond to treatment promptly.