Studies in vitro and in animal models of colorectal and hepatocellular cancers suggest that curcumin is an effective chemopreventive agent. In this pilot trial, we investigated whether oral administration of curcumin results in concentrations of the agent in normal and malignant human liver tissue, which are sufficient to elicit pharmacological activity. In total, 12 patients with hepatic metastases from colorectal cancer received 450 -3600 mg of curcumin daily, for 1 week prior to surgery. Levels of curcumin and its metabolites were measured by HPLC in portal and peripheral blood, bile and liver tissue. Curcumin was poorly available, following oral administration, with low nanomolar levels of the parent compound and its glucuronide and sulphate conjugates found in the peripheral or portal circulation. While curcumin was not found in liver tissue, trace levels of products of its metabolic reduction were detected. In patients who had received curcumin, levels of malondialdehyde-DNA (M 1 G) adduct, which reflect oxidative DNA changes, were not decreased in post-treatment normal and malignant liver tissue when compared to pretreatment samples. The results suggest that doses of curcumin required to furnish hepatic levels sufficient to exert pharmacological activity are probably not feasible in humans.