The clinical value and sensitivity of serum caffeine clearance measurement has been evaluated as an indicator of hepatic disease. After a 17 hour caffeine exclusion period, 300 mg of caffeine citrate was administered orally to the study subjects. Serum samples were taken four and 16 hours later. Serum caffeine concentrations were measured using an enzyme multiplied immunoassay technique (EMIT) and a clearance value derived. Conventional liver function tests were measured at the same time. A total of 103 subjects attending the medical unit in a district general hospital were studied. Twenty one had alcoholic liver disease, 11 non-alcoholic cirrhosis, nine non-cirrhotic liver disease, 21 suspected liver disease, six hepatic tumours, and 35 were hospital and normal control subjects. Caffeine clearance values were lowest in subjects with alcoholic liver disease (median 0.19 ml/min/kg, range 0.04-0.61 ml/min/kg) and significantly reduced in all subjects with liver disease (median 0.32 ml/min/kg, range 0.04-2.68 ml/min/kg) compared with control subjects (median 1.27 ml/min/kg, p less than 0.001). In subjects with suspected liver disease subsequently shown to have another explanation for abnormal liver function test results, caffeine clearance values were normal (median 1.31 ml/min/kg, range 0.23-2.64 ml/min/kg) and significantly different, p less than 0.001, from those of subjects with liver disease. Serum albumen values were not different for these latter two groups. Using a cut off value of 0.86 ml/min/kg, caffeine clearance measurement was 100% sensitive for alcoholic liver disease and 89% sensitive for all liver disease. The respective sensitivities for conventional liver function test measurement were 76% and 83%. In the suspected liver disease group, caffeine clearance was abnormal in only 24%, conventional liver function tests were abnormal in 95%. The respective specificities for caffeine clearance and liver function test measurement in control subjects were 93% and 100%. Caffeine clearance determined by EMIT is a simple inexpensive hepatic metabolic function test. This study indicates that it is a more sensitive indicator of structural liver disease than conventional liver function tests, especially for alcoholic liver disease. The test could be widely introduced as a useful, repeatable assessment of hepatic function.