The aim of this multicentre study was to investigate the effect--in everyday life--of long term administration of acarbose on parameters of glycaemic control, daily insulin requirements, lipid parameters and tolerability in ambulant type 1 diabetic subjects insufficiently controlled with diet and insulin. Furthermore, effects on lipid parameters were to be studied. A total of 16 patients withdrew from the study, 13 of these during the acarbose medication period. For four of these 13 patients the adverse event started during the placebo run-in period. The data of 62 patients (35 men and 27 women, mean age 38 (range 18-64) years, median duration of diabetes 10 (range 1-40) years) were valid for statistical analysis. The median daily dose of acarbose at the final assessment (i.e. after 16 weeks of active treatment) was 200 (range 75-300) mg. During the placebo run-in period HbA1c levels tended to decrease from 8.9 +/- 1.1 to 8.5 +/- 0.9%. After 8 and 16 weeks of acarbose treatment the mean level had decreased further to 8.1 +/- 0.9 and 8.2 +/- 0.9%, respectively (both P < 0.001). After stopping acarbose HbA1c levels increased again to a mean level of 8.6 +/- 0.9%. Mean levels of HbA1c per centre followed the same profile. Seven-point blood glucose profiles followed the same pattern. None of these changes over time reached statistical significance except for a significant drop during acarbose treatment of the time-point 90 min after lunch (P < 0.01). After stopping acarbose treatment values returned to pre-study levels. For total cholesterol, HDL-cholesterol, triglycerides, Apo A1 and Apo B, and Lp(a) no significant changes were observed. Daily insulin dose was 48 (range 26-92) U at the start of the study and did not change. The most frequent reported adverse events were flatulence (43%), diarrhoea (27%), and abdominal pain (11%). We conclude that acarbose up to 3 x 100 mg/day can be a valuable adjunct to insulin in improving metabolic control in persons with type 1 diabetes.