In an attempt to improve tumor response and survival among patients with colorectal cancer metastases confined to the liver, we developed an experimental (rats and pigs) and clinical isolated hepatic perfusion (IHP) technique to exploit maximally the steep dose-response relation of may anticancer drugs. In this overview we present our experimental and clinical results with mitomycin C (MMC) and melphalan (L-Pam). In rats, treatment with a four times higher maximally tolerated dose (MTD) of MMC during IHP compared to hepatic artery infusion (HAI) resulted in higher, more effective intratumoral concentrations of MMC. As a result, only in the IHP-treated rats were complete remissions observed and long-term survival achieved. Hepatotoxic side effects were minimal and transient in all animals. In the clinical phase I/II study with MMC 30 mg/m2 administered as a bolus in the isolated circuit, two of nine patients had a complete remission, with a median survival of 17 months. Four patients developed venoocclusive disease (VOD) of the liver, and as a result one patient died. Therefore we consider MMC unsuitable for further IHP studies. Meanwhile experiments in rats showed that IHP with the L-PAM MTD of 12 mg/kg was even more effective than MMC and did not cause hepatotoxic side effects. In the phase I/II dose-findings study with L-PAM in IHP, the MTD in humans was approximately 3.0 mg/kg. As in the rats, systemic toxicity was dose-limiting. The median survival of the whole group was 18 months. We have started a phase II study of L-PAM in IHP with a fixed dose of 200 mg L-PAM to determine if IHP can significantly increase the median survival and complete remission rate compared to other treatment modalities. Results from this study are expected by the end of 1997.