Three muscle biopsies were performed in 53 overt type II diabetics over a period of approximately 2 years. At baseline, 21 (40%) had an abnormally increased capillary basement membrane width in muscle. Thirty-five subjects received glipizide and 18, placebo. At baseline, no statistically significant difference was found in the muscle capillary basement membrane width between the two groups (P = NS). In the subjects receiving placebo, the mean width of the muscle capillary basement membrane increased (P = NS), but in those receiving glipizide, the mean decreased from 193 +/- 13 nm (SEM) to 161 +/- 10 nm (P = .02). Fasting plasma glucose and glycosylated hemoglobin A1 significantly decreased (P less than .001) after two years in those receiving glipizide. In 15 subjects, mean glycosylated hemoglobin A1 reached the normal range, and mean muscle capillary basement membrane width decreased to a level close to that found in subjects without diabetes (P = NS). Determinations of enzyme activities involved in the synthesis and degradation of glycoproteins revealed a 2-year significant decrease of muscle glucosyltransferase (synthesis) activity (P less than .01) in the glipizide-treated subjects as opposed to a significant increase (P less than .001) in those receiving placebo. Muscle N-acetyl-beta-glucosaminidase activity (degradation) was statistically increased (P less than .001) in those subjects taking glipizide, but decreased in those taking placebo (P less than .001).