The administration of metformin, as glucophage retard, at bedtime instead of supper time may improve diabetes control by reducing morning hyperglycemia. This modification of glucophage treatment was tried in 3 groups of diabetic patients: I. those with secondary failure of routine treatment with sulfonylurea (SU) and glucophage; II. those with combined SU and bedtime insulin; III. Type 1 patients with early morning hypoglycemia. The first 3 months of observation in 258 patients showed that 136 (52.7%) reacted very well to the change. In Group I the addition of insulin to SU could be postponed. In Group II, night insulin could be reduced or eliminated. In Group III, evening or night insulin could be reduced by up to 70%. There was no early morning hypoglycemia nor morning hyperglycemia. The success rate in the 2 Type 2 groups was better (72% and 60%) than in the Type 1 group (34%). 30 patients (11.6%) had to stop the treatment because of side effects of the glucophage (mainly diarrhea or nausea). So far, we have found no clinical signs that might indicate which patients might benefit from this modification of treatment. A fasting blood sugar done within 2-3 days after the change in treatment may immediately indicate whether the new treatment is effective.