Clozapine is a potent antipsychotic agent that has been marketed since 1990. Several published reports of diabetes mellitus occurring with clozapine therapy have appeared during the past 5 years. Because the risk and characteristics of clozapine-associated diabetes mellitus remain unclear, we conducted a descriptive epidemiologic study of spontaneous adverse event reports of hyperglycemia occurring in clozapine-treated patients. The Food and Drug Administration MedWatch surveillance program was queried (January 1990 through February 2001), and the results were pooled with published cases. Parameters assessed included documentation of diabetes, clinical severity, new-onset diabetes versus exacerbation of preexisting disease, demographic characteristics of patients, time to onset of hyperglycemia, and effect of drug discontinuation and rechallenge. We identified 384 reports. Of these, new-onset diabetes was diagnosed definitively in 242 patients, and 54 patients had exacerbation of preexisting disease. The mean (+/- SD) age was 40 +/- 12 years (range, 13 to 77). The male:female ratio was 2:0. Most cases appeared within 6 months of initiating clozapine therapy. One patient developed diabetes following a single 500-mg dose. There were 80 cases of metabolic acidosis or ketosis. Twenty-five patients died during hyperglycemic episodes. Forty-six patients had improved glycemic control after discontinuation or dose reduction of the drug.A causal relationship between clozapine and diabetes is suggested by the number of reports, the temporal relation to clozapine initiation, the relatively young age of the affected patients, and the prompt reversibility on withdrawal of the drug in some patients. The severity of reported cases ranged from mild glucose intolerance to diabetic ketoacidosis or hyperosmolar coma.