Serum concentrations of caffeine (CA) and its three major dimethylmetabolites (theobromine; TB: paraxanthine; PX: theophylline; TP) were measured in fifteen patients with cholelithiasis, in ten patients with cirrhosis and in ten healthy subjects after the oral CA (2 mg/kg) loading. The correlations of total body clearance (CL) between three-point study (sampling times 1, 2 and 4 h) and nine-point study (sampling times 0.5, 1, 2, 4, 6, 8, 10, 12 and 24 h) were highly significantly correlated (r = 0.988, p less than 0.001). The elimination half-life (t1/2) of CA was significantly longer in cirrhotic patients than in the other two groups. Cirrhosis had no effect on the apparent volume of distribution (Vd) of CA, but CL of CA was substantially reduced in these patients. Production of the three metabolites of CA, but mainly PX, was reduced in patients with cirrhosis. There were significant correlations between the serum PX/CA (r = 0.911, p less than 0.001) and (PX + TB + TP)/CA (r = 0.905, p less than 0.001) ratios and CL of CA at 4 h after CA administration in the three groups. These findings suggest that CA pharmacokinetic parameters can be estimated using a simplified three-point blood sampling procedure following a single oral load and that the serum PX/CA or (PX + TB + TP)/CA ratio in a single blood sample taken 2 or 4 h after dosing provides a useful indicator for the assessment of hepatic drug-oxidizing capacity, N-demethylation, in decompensated liver cirrhosis. However, CA test was unable to distinguish the difference of liver function between the control subjects and in patients with cholelithiasis.