The cost-effectiveness component of the 18-month CATIE trial of schizophrenia pharmacotherapy (n = 1460) showed that the first-generation antipsychotic perphenazine was US$300-600 per month less expensive than each of four second-generation antipsychotics, and no less effective across multiple measures. We consider whether or not each of eight potential methodological limitations could weaken this conclusion: follow-up rates, study duration, sample characteristics, the choice of outcome measures, exclusion of patients with tardive dyskinesia from assignment to perphenazine, choice of study drugs and doses, reliance on intention-to-treat analysis, and differences in prestudy treatment. We conclude that results of CATIE are robust to these limitations. Perphenazine seems to have been a more representative choice for first-generation antipsychotic comparison treatment than haloperidol.