Forty-four fatal cases of Rocky Mountain spotted fever (RMSF) occurring in 1974 were compared with 50 nonfatal cases of similar age, sex, date of onset, and place of occurrence. Diagnosis and initiation of treatment in fatal cases were substantially delayed compared with nonfatal cases. Several reasons for this delay were identified: (1) the rash appeared later in the course of illness in the fatal cases, often not until the patient was terminal, (2) a history of tick bite was less often obtained during life or obtained late in the clinical course in fatal cases, and (3) initial nonspecific symptoms or unexpected symptoms led to an initial diagnosis of more common diseases. Only two fatal cases were treated with either tetracycline or chloramphenicol before the sixth day of illness. Presumptive diagnosis of RMSF and initiation of tetracycline therapy before onset of rash may be necessary to reduce mortality.