Pharmacokinetics and dose-finding studies on sorafenib were conducted on heterogeneous groups of patients with solid tumors. Portal hypertension, gut motility impairment and altered bile enterohepatic circulation may explain different sorafenib toxicological profile in cirrhotic patients. This study evaluated sorafenib plasma concentration in a homogeneous group of cirrhotic patients with hepatocellular carcinoma (HCC). Sorafenib concentrations were determined by liquid chromatography in 12 consecutive patients. Data have been evaluated by the generalized estimating equations method (p value statistical level was set at α = 0.05). (1) There were not significant differences between sorafenib concentrations in patients who tolerate the full dose versus patients with reduced dose due to toxicity; (2) the average sorafenib concentrations measured 3 h after the morning dosing were lower than those measured 12 h after the evening dosing (p = 0.005); (3) sorafenib concentrations decrease overtime (p < 10(-4)); (4) it has been found an association between the development of severe adverse reactions and sorafenib concentrations (p < 10(-5)). The relationship between dose and concentration of sorafenib in HCC patients is poor and not clinically predictable, confirming the variability both in the maximum tolerated dose and in plasma concentrations. Several factors may influence the pharmacokinetics in patients with liver disease. This may explain the inter-patient variability of concentrations and the lack of differences in concentration at different dosages. It could be interesting to extend the series of HCC patients to enhance information on the kinetics of the drug; furthermore, to establish a threshold of plasma sorafenib concentrations to predict severe adverse reactions would be clinically useful.