It has been suggested that clozapine has one of the largest diabetic effects of all atypical antipsychotics. To confirm these findings, we examined retrospectively the risk of new-onset diabetes in long-term clozapine treatment compared to treatment with other antipsychotics in a matched control population with schizophrenia or schizoaffective disorder. Ninety-four adult patients with schizophrenia or schizoaffective disorder who had been treated with clozapine for 5 years or longer were matched on age, diagnosis, and sex to 94 patients without any use of clozapine. The groups were followed up for as long as 20 years. The cumulative incidence of new detection of diabetes in the clozapine group was 22.3% (mean follow-up, 12.3 years; absolute risk difference, 6.3%; 95% confidence interval, -4.9% to 17.5%). An additional rigorous analysis of the 83 matched pairs with normal glucose measurement before end point showed a significant risk difference between the 2 groups (21.7% compared with 8.4%) but may have been biased against clozapine. We conclude that definitive evidence showing a clinically significant larger risk for new-onset diabetes after long-term treatment with clozapine in comparison to other antipsychotics is lacking.