Epidemiologic and clinical characteristics of fatal and nonfatal cases of Rocky Mountain spotted fever (RMSF) were compared to identify risk factors for death caused by this disease. Confirmed and probable RMSF cases reported through US national surveillance for 1981-1998 were analyzed. Among 6388 RMSF patients, 213 died (annual case-fatality rate, 3.3%; range, 4.9% in 1982 to 1.1% in 1996). Use of tetracycline-class antibiotics for treatment of RMSF increased significantly in the 1990s, compared with use in the 1980s. Older patients, patients treated with chloramphenicol only, patients for whom tetracycline antibiotics were not the primary therapy, and patients for whom treatment was delayed > or =5 days after the onset of symptoms were at higher risk for death. Although the case-fatality rate was lower in the 1990s than in the 1980s, risk factors for fatal RMSF were similar. Despite the availability of effective antibiotics, RMSF-related deaths continue to occur because of delayed diagnosis and failure to use appropriate therapy.