A recently developed chromogenic endotoxin assay with plasma Tween 80 pretreatment was compared with the conventional dilution and heating method. Plasma endotoxin was measured in patients with liver cirrhosis and upper gastrointestinal (GI) bleeding by these methods. Plasma endotoxin concentration was calculated from an individual internal standard curve by adding three different standard endotoxin solutions to each sample. By the conventional heating method, added standard endotoxin gave different OD values in each sample and the slope of the standard curve showed interindividual variations. When sample plasma from chronic alcoholics was pretreated with 1% Tween 80 and ultrasonification after heating, the slope of standard curves was somewhat increased and interindividual variation was minimized. Significantly higher plasma endotoxin levels in cirrhotics with upper GI bleeding compared with those without upper GI bleeding was detected by this Tween 80 method. There was a strongly positive correlation between the endotoxin levels determined by this method and those determined by the perchloric acid method and endotoxin-specific substrate in patients with upper GI bleeding. Endotoxin levels, which were elevated 1-2 days after the bleeding, tended to decrease as patients recovered. In summary, the recovery of endogenous and exogenous endotoxin from plasma sample was increased by adding Tween 80 before the chromogenic substrate assay. Transient elevation of plasma endotoxin was demonstrated by this Tween 80 method in patients with liver cirrhosis and upper GI bleeding.