To investigate the specificity of the dexamethasone suppression test (DST) for the diagnosis of major depression in patients with diabetes mellitus, we administered 1 mg of dexamethasone to 30 nondepressed diabetics and to 58 normal controls at 11 PM. Diabetic subjects received hemoglobin A1 (Hb A1) determinations, the Hamilton Rating Scale for Depression (HRSD), and five to eight blood glucose determinations during the 48 hours surrounding the DST. Results demonstrated a significantly higher rate of nonsuppression (plasma cortisol level, greater than or equal to 5 micrograms/dL) at 4 PM the following day among diabetics (43%) than among controls (7%) but no difference between these groups in the rate of nonsuppression at 8 AM. Plasma cortisol level at 4 PM correlated with Hb A1 level but not with duration of illness, HRSD score, mean blood glucose level, or maximum blood glucose excursion. These results suggest that the results of the DST used as a diagnostic test for major depression must be interpreted with caution in patients with diabetes.