Nine patients (6 males, 3 females) with biopsy-proven liver cirrhosis participated in an open, cross-over, three centre study of the effect of impaired liver function on the pharmacokinetics of felodipine. Two of the nine patients had undergone porto-caval anastomosis. Each patient was given 0.75 mg i.v. and 10 mg p.o. on separate occasions. The results of this study have been compared with published data from younger subjects and elderly hypertensive patients. The mean peak plasma concentration normalized to a dose of 10 mg (Cmax 46 nmol/l) was twice as high in the cirrhotic patients as in the healthy subjects, but the bioavailability, f, (17.0%) was comparable. Subjects with a porto-caval shunt did not have higher f than the mean for the group. The volume of distribution at steady-state, Vss, was significantly lower than in the healthy subjects. Protein binding was significantly lower in the patients with cirrhosis: 99.46% compared to 99.64% in the healthy subjects. The weight-corrected clearance was 1/3 of the value in healthy subjects. No correlation between systemic availability and oral clearance was found, so it is proposed that felodipine is metabolized both in the liver and also in the gut wall. The results suggest that at least the starting dose should be reduced in patients with severe liver disease.