Trilostane and aminoglutethimide, both given with a physiological replacement dose of hydrocortisone, were randomly allocated to 112 eligible patients with postmenopausal advanced breast cancer. Following treatment failure on either drug the patient continued with the other, if they were in a suitable clinical condition. Sixty-three patients initially received trilostane, of whom 33 subsequently received aminoglutethimide; 49 patients initially had aminoglutethimide and 14 of these then received trilostane. Both groups of patients were comparable in all respects. There was no difference in the response rate to either drug or in the average time to disease progression for the two drugs. Of the 47 patients who received both drugs, nine (19%) showed a response to both, indicating no cross-resistance. Side effects were seen to both drugs in approximately half the patients; these were mainly gastrointestinal symptoms with trilostane and rashes and drowsiness with aminoglutethimide. There was no evidence of cross-over patient susceptibility to side effects.