Background: We hypothesized that acarbose would delay conversion from impaired glucose tolerance (IGT) to type 2 diabetes by alleviating postprandial hyperglycaemia. Our study's main objective was to investigate the effect of acarbose in IGT-persons on their 2-h plasma glucose level and beta-cell function. Subjects and Methods: The study included a random sample of 45-70-year-old residents of Hoorn, Netherlands, with mean fasting plasma glucose <7.8 mmol/L and mean 2-h plasma glucose of 8.6-11.1 mmol/L (measured by two successive oral glucose tolerance tests). After a qualification period, participants were randomized to acarbose treatment or placebo. Insulin secretion and insulin sensitivity were measured by hyperglycaemic clamp. After a 3-year treatment, analyses were performed of both the intention-to-treat and the per-protocol groups. Results: Of the 12 093 residents who received postal invitations, 118 participants were randomized. The mean difference of the post-load plasma glucose after 3 years, was -1.16 mmol/L (95% CI: -2.03; -0.17). The absolute risk reduction for diabetes was 6% (95% CI: -9; 21). No effect was seen on insulin secretion and insulin sensitivity. Conclusions: In patients with IGT, treatment with acarbose was associated with beneficial effects on 2-h plasma glucose levels but not with improvement of beta-cell function. Copyright © 2008 John Wiley & Sons, Ltd.