BACKGROUND The 13C-galactose breath test has been reported to be an accurate, non-invasive method for the assessment of liver function. OBJECTIVE To determine the optimal doses of labelled and unlabelled carrier galactose necessary to perform the 13C-galactose breath test, to assess the utility of the 13C-galactose breath test in distinguishing between normal subjects and those with liver cirrhosis and to determine whether the 13C-galactose breath test can stratify patients with cirrhosis based on their Child-Pugh score. METHODS Twenty-three control subjects and 30 patients with liver cirrhosis received fixed doses of unlabelled carrier galactose and labelled 13C-galactose. Breath samples were collected just before and at 30-min intervals up to 4 h after the ingestion of unlabelled carrier galactose and labelled 13C-galactose. Each sample was analysed for its 13CO2 content. RESULTS Doses of 25 g/m2 of unlabelled carrier galactose and 100 mg of 13C-galactose had the greatest sensitivity (93%; 95% confidence interval, 76-99%) and specificity (87%; 95% confidence interval, 65-97%) for distinguishing between normal subjects and cirrhotics when the test was performed 2 h after ingestion. The 13C-galactose breath test was also able to distinguish between class A and class B or C cirrhotics. CONCLUSIONS The 13C-galactose breath test is a useful non-invasive tool for distinguishing between healthy subjects and patients with liver cirrhosis and between cirrhotics with well-compensated liver disease and those with decompensated liver disease.