PURPOSE: To compare the therapeutic potential of acarbose, metformin, or placebo as first line treatment in patients with non-insulin-dependent diabetes mellitus (NIDDM). PATIENTS AND METHODS: Ninety-six patients with NIDDM (35-70 years of age, body mass index (BMI) [is less than or equal to] 35 kg/[m.sub.2], insufficiently treated with diet alone, glycated hemoglobin ([HbA.sub.1c]; 7% to 11%) were randomized into 3 groups and treated for 24 weeks with acarbose, 3 x 100 mg/day, or metformin, 2 x 850 mg/day, or placebo. Efficacy, based on [Hba.sub.1c] (primary efficacy criterion), fasting blood glucose (BG) and insulin, 1 hour postprandial BG and insulin (after standard meal test), postprandial insulin increase, plasma lipid profile, and tolerability, based on subjective symptoms and laboratory values were determined every 6 weeks. Analysis of covariance was performed for endvalues with adjustment on baseline values. Ninety-four patients were valid for efficacy evaluation. RESULTS: Both active drugs showed the same improvement of efficacy criteria compared with placebo. Baseline adjusted means at endpoint were as follows: BG, fasting and 1 hour postprandial, 9.2 mM and 10.9 mM with placebo, 7.6 mM and 8.7 mM with acarbose, and 7.8 mM and 9.0 mM with metformin; [Hba.sub.1c] was 9.8% with placebo, 8.5% with acarbose, and 8.7% with metformin. Comparisons: acarbose versus placebo and metformin versus placebo were statistically significant, but not acarbose versus metformin. No effect on fasting insulin could be observed. Relative postprandial insulin increase was 1.90 with placebo, 1.09 with acarbose, and 1.03 with metformin. Comparisons: acarbose versus placebo and metformin versus placebo were statistically significant, but not acarbose versus metformin. With respect to lipid profile, acarbose was superior to metformin. Low-density lipoprotein (LdL)/high-density lipoprotein (HDL) cholesterol ratio increased by 14.4% with placebo, was unchanged with metformin, but decreased by 26.7% with acarbose. Comparisons: acarbose versus placebo and acarbose versus metformin were statistically significant, but not metformin versus placebo. Slight body weight changes were observed with acarbose (-O.8 kg) and metformin (-0.5 kg), but not with placebo. Acarbose led to mild or moderate intestinal symptoms in 50% of the patients within the first 4 weeks, but in only 13.8% of the patients within the last 4 weeks. CONCLUSIONS: Acarbose and metformin are effective drugs for the first line monotherapy of patients with NIDDM. With respect to plasma lipid profile, especially HDL cholesterol, LDL cholesterol and LDL/HDL cholesterol ratio, acarbose may be superior to metformin.