OBJECTIVE: To determine the incidence and timing of relapses in patients who have recovered from an acute episode of thrombotic thrombocytopenic purpura. DESIGN: Clinical follow-up for 3 to 10 years. SETTING: General community outpatient study; patients who had relapse were hospitalized. PARTICIPANTS: 63 of 72 surviving patients who had participated in a randomized study that compared plasma exchange and plasma infusion as treatments for thrombotic thrombocytopenic purpura and for whom continued clinical follow-up was obtained. OUTCOME MEASURES: Recurrence of thrombotic thrombocytopenic purpura as defined by a decrease in platelet count to less than 100 x 10(9)/L and by the onset of microangiopathic hemolytic anemia as identified by erythrocyte fragmentation in a peripheral blood film. RESULTS: 37 of the 63 patients have not had recurrence of thrombotic thrombocytopenic purpura and have remained completely well; 6 patients have not had recurrence but have developed other medical problems; 3 patients have not had recurrence but have residual neurologic defects from the original episode; and 17 patients have had one or more recurrences, occurring 7 months to 8 years after the original episode. As determined by Kaplan-Meier analysis, the projected recurrence rate after 10 years in all surviving patients is 36% (95% CI, 23% to 59%). CONCLUSIONS: More than one third of patients who survive an acute episode of thrombotic thrombocytopenic purpura will have at least one relapse during the following 10 years.