This study investigated the long-term effect of insulin or the combination of insulin and an oral hypoglycemic compound (glipizide) on the skeletal muscle capillary basement membrane width in insulin-requiring diabetic patients. Seventy diabetic patients were randomized to treatment with either insulin-placebo or insulin-glipizide (5 mg/d) for 3 years. Of these, only 61 patients completed the study; 27 patients received insulin-placebo and 34 patients received insulin-glipizide. Three skeletal muscle (quadriceps femoris) biopsies were performed in all patients over a 3-year period. Glycosylated hemoglobin A1 was determined every 100 +/- 20 days, including plasma glucose levels. Muscle capillary basement membrane width was quantitated by a previously described method. After approximately 16 months, glycosylated hemoglobin A1 decreased significantly in each group from its baseline (P < 0.001 insulin-glipizide group and P < 0.025 insulin-placebo), although no statistically significant difference was seen between the two groups. After 3 years this decrease was statistically significant (P < 0.001) only in the insulin-glipizide group. At baseline, no statistically significant difference was found in the muscle capillary basement membrane width between the two groups. In spite of the significant decrease in glycosylated hemoglobin A1 in both groups after 14 to 16 months, only muscle capillary basement membrane width in the insulin-glipizide group decreased significantly compared with baseline. Patients receiving insulin-placebo showed a gradual increase in the muscle capillary basement membrane width, which after 3 years was significantly higher than baseline (P < 0.02). Although the mechanisms by which the addition of glipizide to insulin treatment reduced the thickening of the muscle capillary basement membrane are not clearly understood, the current findings suggest that diabetic microangiopathy is not necessarily progressive and that prophylaxis may be attained.