We attempted to define the most effective treatment for polyarteritis nodosa and Churg-Strauss angiitis, with a prospective, randomized, multicenter trial of cyclophosphamide in conjunction with corticosteroids and plasma exchanges, compared to corticosteroids and plasma exchanges. A total of 71 patients who fulfilled clinical, histological and/or arteriographic diagnostic criteria were randomly designated to receive either prednisone and plasma exchanges (group A, n = 39) or cyclophosphamide, prednisone and plasma exchanges (group B, n = 32). The end points of the study were control of the disease (recovery and remission) and death. Upon study entry clinical and laboratory features did not differ in the 2 groups. Treatment was stopped in 19 patients because of ineffectiveness in 10 (9 in Group A) and side effects in 9 (8 in Group B). Initial control of the disease was similar in both groups. At 5 years, 27 patients had completely recovered and 14 patients were in clinical remission. The cyclophosphamide-prednisone-plasma exchange association was beneficial in preventing relapses during longterm followup. Nineteen deaths were reported during the followup period. There was no difference between the 10 year cumulative survival rates of the 2 groups (respectively, 72 and 75%). Thus, the association of cyclophosphamide with corticosteroids and plasma exchanges reduced the incidence of relapses and improved the quality of the clinical response to therapy.