The course of depression in patients with comorbid medical illness is poorly understood. We report a 5-year follow-up study of 25 diabetic patients who had participated in an 8-week depression treatment trial. When a patient completed the trial, primary physicians were informed of patient outcomes and advised to monitor for relapse and treat those with ongoing depression. At the 5-year reevaluation depression was assessed using DSM-III-R criteria, and a depression severity scale was formed that reflected the presence, severity, frequency, and duration of depression episodes as well as a global assessment of functioning. Recurrence or persistence of depression occurred in 23 (92%) of the patients with an average of 4.8 depression episodes over the 5-year follow-up period. The duration of the longest episode averaged 16 +/- 4 months. Reversion to major depression occurred frequently and rapidly also in the subset that remitted during the treatment trial: 58.3% were depressed again within the first year. At the time of the follow-up interview, major depression was evident in 16 (64%) of the subjects, and glycemic control was significantly worse in this group compared with those without depression (gHb: 13.3% +/- 2.6% vs 11.1% +/- 1.9%, P = 0.03). Severity of depression over follow-up was related to the presence of neuropathy at entry and to incomplete remission during the initial treatment trial. Nineteen patients (82.6% of those who relapsed) received additional courses of antidepressant therapy, but none was treated continuously for depression prophylaxis. In this diabetic sample, depression was a recurrent condition in the vast majority of cases, and initial treatment response did not confer lasting euthymia. Whether maintenance antidepressant medication would be useful in preventing depression recurrence and promoting better glycemic control in diabetes remains to be studied.