The lifetime prevalence of psychiatric illness was determined in 114 patients with diabetes mellitus (types I and II) who were selected randomly from patients undergoing diabetes evaluations at a large medical center. The relationship of psychiatric illness to diabetic control was studied using glycosylated hemoglobin (HbA1) and self-report measures of metabolic symptoms. Seventy-one percent of the patients had a lifetime history of at least one criteria-defined psychiatric illness; affective and anxiety disorders were the most common diagnoses. A significant difference (p = .02) in mean glycosylated hemoglobin levels was observed comparing patients with a recent psychiatric illness (mean = 10.8%) to those never psychiatrically ill (mean = 9.6%). These psychiatrically ill patients also reported more symptoms of poor metabolic control and more distress associated with these symptoms than did patients never psychiatrically ill (p less than .0001 for both). The overall report of diabetes symptoms was unrelated to HbA1 (p = .25) and was influenced primarily by the recent presence of psychiatric disorder (p less than .0001). We conclude that emotional illness is associated with both poorer metabolic control and the increased report of clinical symptoms of diabetes.