We carried out a double-blind, placebo-controlled cross-over trial to evaluate changes in nutrient intake during acarbose therapy and short-term changes in metabolic control in Finnish patients with non-insulin dependent diabetes mellitus. The efficacy and tolerability of acarbose was also assessed during this short 8-week treatment period. Patients were treated either with diet only (18 subjects) or with diet and sulphonylurea (9 subjects) prior to entering this trial. The nutrient intake was assessed using food records kept during the last 2-5 days of each treatment period. Treatment with acarbose did not affect body weight or the intake of energy and carbohydrates, but it was associated with a slight decrease in the dietary fat intake. Both fasting and 2-h postprandial blood glucose levels fell continuously during acarbose but the change did not reach statistical significance during the 8-week treatment period. The 2-h serum insulin response was not affected. The results were similar in patients treated with acarbose alone and in those receiving acarbose together with sulphonylurea regimen. The acarbose regimen was relatively well tolerated although mild gastrointestinal side-effects were observed in many patients. In conclusion, treatment with acarbose has no major effects on nutrient intake in diabetic patients. To monitor changes in metabolic control after the introduction of acarbose to the treatment regimen a longer than 8-week period is needed.