Non-insulin-dependent diabetes mellitus not responding to diet only in patients with non-alcoholic liver cirrhosis is characterized by high post-prandial hyperglycemia. The aim of this study was to evaluate the safety and efficacy of 24 weeks of treatment with 300 mg acarbose per day in 76 consecutive outpatients affected by type 2 diabetes and well-compensated liver cirrhosis. The study design was double-blind cross-over vs placebo. All patients tolerated both treatments well, and no significant variations in liver function tests were observed (< 5% vs pre-treatment). A significant reduction of several parameters was observed only after acarbose: fasting glycemia (19 +/- 6 vs 2 +/- 0.5%; p < 0.01), post-prandial glycemia (41 +/- 9 vs 3 +/- 0.6%; p < 0.01), mean glycemia (30 +/- 8 vs 14 +/- 5%; p < 0.01), daily glycemic variation (52 +/- 8 vs 8 +/- 1%; p < 0.01), HbA1c (16 +/- 1 vs 2 +/- 0.5; p < 0.05), incremental area of C-peptide after a standard meal (80 +/- 19 vs 200 +/- 36 ng/mL/300 min; p < 0.01). After acarbose a significant increase of intestinal voiding/week (98 vs 28%; p < 0.01) and a parallel reduction of blood ammonia levels (52 +/- 9 vs 9 +/- 5%; p < 0.01) were observed. Results clearly document the good tolerability and the absence of toxic effects of acarbose on the liver, due to a theoretic absence of both absorption by the gut and hepatic metabolism of the drug. In fact, acarbose increases peristaltic movement of the gut, stimulates the proliferation of saccharolytic bacteria and simultaneously reduces proteolytic bacterial proliferation, thus actively reducing blood ammonia levels. These unexpected effects of acarbose may be used to advantage for the treatment of type 2 diabetes mellitus in patients with well-compensated liver cirrhosis.