To evaluate the clinical usefulness of temporal artery biopsy in the diagnosis of giant-cell arteritis we followed up all 134 residents of Olmsted County, Minnesota, who had temporal artery biopsies between 1965 and 1980. Initial biopsies were positive for giant-cell arteritis in 46 cases and negative in 88. A history of jaw pain or claudication and the findings of a palpably abnormal temporal artery were significantly more common in the patients whose biopsy specimens showed giant-cell arteritis. Over a median follow-up period of 70 months (range 1-192) only 8 of the 88 biopsy-negative patients had clinical courses requiring long-term, high-dose corticosteroid therapy for giant-cell arteritis. In this population-based study temporal artery biopsy correctly predicted the subsequent need for corticosteroid therapy in 94% of cases: these findings indicate that biopsy should be done before patients are committed to long-term corticosteroid therapy.