The effect of a good metabolic control in the natural history of diabetic retinopathy is discussed because, although the onset and the incidence are closely related to a bad metabolic control, there is no improvement related to a good control; progression having been reported with the good metabolic controls. 7 patients subjected to a strict control with continuous subcutaneous insulin infusion were compared with a group of 10 patients with a good conventional treatment in an 18 month study. The best metabolic control of the first group was obtained at 9, 12, 15 and 18 months (p less than 0,05). The evolution of retinopathy in this group did not change with the treatment and there was no proof of progression with the use of continuous subcutaneous infusion. Further follow-up evaluation is necessary to confirm this affirmation.