Plasma endotoxin concentration was measured in 85 patients with alcoholic liver disease (alcoholic cirrhosis (n = 64), alcoholic hepatitis without cirrhosis (n = 11), fatty liver (n = 10), and in patients with non-alcoholic cirrhosis (n = 15]. Endotoxin concentration was determined with an improved chromogenic substrate assay, using individual standard curves for each plasma sample. In patients with alcoholic cirrhosis the mean endotoxin concentration was significantly higher than in patients with non-alcoholic cirrhosis (p less than 0.05). In addition, distinctly higher endotoxin concentrations (greater than 20 pg/ml) were more frequently observed in patients with alcoholic cirrhosis than in non-alcoholic cirrhosis (34.4 vs. 14.3%, p less than 0.05). Mean endotoxin concentration was not significantly higher in cirrhotics with ascites or esophageal varices as compared with the subgroup without ascites or esophageal varices. The endotoxin concentration did not correlate with serum bilirubin, prothrombin concentration or serum enzyme activities. In patients with alcoholic liver disease, however, endotoxin concentration revealed a negative correlation (p less than 0.05) with the concentration of high density lipoprotein cholesterol. On admission endotoxin concentrations in alcoholics with fatty liver were similarly elevated as observed in alcoholic cirrhosis. In six out of 12 patients with fatty liver or alcoholic hepatitis, in whom a second sample of plasma was investigated after 6 to 8 days, endotoxemia was no longer detectable; in the remaining patients, the endotoxin concentration decreased markedly. The results indicate that, irrespective of the stage of liver disease, alcohol abuse favours the development of endotoxemia. They support the hypothesis that gut-derived endotoxins might play a role in the initiation and aggravation of alcohol-induced liver disease.