The feasibility of measuring caffeine clearance from saliva (SCI) was assessed in ambulatory patients with liver disease and in a control group, and the results were compared with quantitative liver function tests. For this purpose, the subjects were given 280 mg caffeine p.o. in decaffeinated coffee powder between noon and 4 p.m., and caffeine concentrations were measured in saliva (using an enzyme immunoassay) before bedtime and upon arising. In the cirrhotics (n = 29), SCI was 0.58 ± S.D. 0.45 ml per min × kg, thus being reduced to approximately one-third of drug-free, nonsmoking controls (1.53 ± 0.46, n = 18); although patients with non-cirrhotic liver disease showed intermediate values (0.95 ± 0.47), their reduction in SCI was significant (p < 0.001). SCI was correlated with indocyanine green fractional clearance, galactose elimination capacity and aminopyrine breath test; however, the closest relationship (Rs = 0.80) was observed with the aminopyrine breath test. It is suggested that the measurement of SCI represents a noninvasive and innocuous procedure for quantifying hepatic microsomal function, and is suitable for routine use. Since a.m. saliva concentrations of caffeine are highly correlated (Rs = −0.94) with SCl, further simplification of the test to a single-point measurement appears possible.