Acarbose, an alpha-glucosidase inhibitor, is a new antihyperglycaemic agent which has been proposed as add-on therapy in Type 2 diabetic patients not well-controlled with diet alone, sulphonylurea, metformin or insulin, and in Type 1 diabetic patients with large meal-related plasma glucose excursions. Numerous controlled studies investigating the clinical effects of acarbose in Type 2 diabetes versus either placebo or, more rarely, versus a reference drug (sulphonylurea or metformin) have been published during the last 10 years. All placebo-controlled studies have demonstrated the superiority of acarbose, at a dose of 150-600 mg/day, in decreasing fasting and postprandial glucose levels as well as HbA1c concentrations (mean decrease of 0.7%), whether acarbose was given as first-line therapy in diet-treated diabetic patients or in combination in individuals already receiving a sulphonylurea, metformin or insulin. Only a few controlled studies have compared the effects of acarbose with those of either sulphonylurea or metformin, yielding controversial results. In Type 1 diabetic patients, a small reduction of HbA1c levels was also reported after addition of acarbose to insulin therapy, which in some cases allowed a slight reduction of daily insulin needs. All these favourable biological effects occurred without exposing the patient to hypoglycaemia or weight gain. A few studies have also reported favourable effects on postprandial lipid profile and some other vascular risk factors. However, it is not clear whether the extra cost of acarbose, when compared to that of older oral antidiabetic agents, is justified since no study has yet demonstrated its potential benefit on the complications and long-term prognosis of diabetic patients.