Twenty-two patients with connective tissue diseases, including 14 with systemic lupus erythematosus, were randomly assigned to either prednisone alone or cyclophosphamide (Cytoxan) alone treatment groups. Patients declared a failure on one regimen were then placed in the other group. Initial clinical and laboratory status of the groups was comparable. In the cyclophosphamide groups, zero responses and 12 failures were recorded. With prednisone, 14 responses and 5 failures were observed; similar differences were present within each disease subgroup. Inability to control inflammatory aspects of the disease, or severe side effects, were the usual reason for discontinuing of cyclophosphamide; only rarely was it possible to obtain a prolonged trial on this agent alone. Side effects were distressingly frequent with each drug, and of similar magnitude. Ovarian suppression was documented in several patients in the cyclophosphamide group. Cyclophosphamide employed as the sole agent is not the drug of choice in the treatment of these diseases.