Regional chemotherapy allows further exploitation of the steep dose response curve of most chemotherapeutic agents, while systemic toxicity remains tolerable. We investigated the difference in maximally tolerated dose, pharmacokinetics and antitumour effect comparing administration of melphalan as a bolus in isolated liver perfusion (ILP) or via hepatic artery infusion (HAI). For these in vivo studies an experimental model for liver metastases in male WAG/Ola rats is obtained by subcapsular inoculation of CC531 rat colon carcinoma cells. In this system, ILP allowed administration of a two times higher dose than HAI (12 mg kg-1 vs 6 mg kg-1). In both treatment modalities systemic toxicity (leukopenia) was dose limiting. No hepatic toxicity was observed. Bolus administration of the maximally tolerated doses of melphalan in HAI (6 mg kg-1) and ILP (12 mg kg-1) resulted in four times higher concentrations in both liver and tumour tissue of the ILP treated rats. However, the ratio of mean drug concentration in liver vs tumour tissue appeared to be 1.5 times that found for HAI. In the range of the in tumour tissue measured melphalan concentrations the CC531 cells showed a steep dose response relationship in vitro. Whereas HAI resulted in significant tumour growth delay, complete remissions were observed in 90% of the rats treated with ILP. This study shows that with 12 mg kg-1 melphalan in ILP highly effective drug concentrations are achieved in CC531 tumour tissue; although the melphalan concentration in liver tissue shows an even higher increase than in tumour tissue, hepatic toxicity is negligible in this dose range.(ABSTRACT TRUNCATED AT 250 WORDS)