We evaluated the technical feasibility of isolated liver perfusion (ILP) in the treatment of patients with colorectal cancer metastases confined to the liver, and investigated whether ILP allows exposure of the tumor to high concentrations of mitomycin C (MMC). Furthermore, survival time and tumor response were studied. Nine patients were treated with 30 mg/m2 MMC recirculated for one hour in the isolated circuit. The MMC concentration in the perfusate and plasma was measured using a high-performance liquid chromatography assay. All complications directly related to the surgical procedure were treated effectively (no mortality). The peak concentration of MMC in the perfusate was 5 to 11 times higher than that measured in the plasma of patients treated with 20 to 60 mg/m2 MMC i.v., and the concentration remained significantly higher during the whole perfusion period. In contrast, the peak concentration of MMC in plasma was approximately two thirds of the lowest peak plasma level measured after i.v. administration of 10 mg/m2 MMC. No systemic toxicity was observed in any of our patients. However, four patients developed veno-occlusive disease of the liver which was mild in three but lethal in one. One of the eight evaluable patients had an objective complete response (25 months), one an objective partial response and five others a clear reduction in tumor size (25-50%). The median survival time was 17 months. This study demonstrates that ILP is technically feasible in patients, and in comparison with systemic therapy allows exposure of hepatic metastases to much higher concentrations of MMC, while systemic toxicity is absent. Remarkably, this single exposure to a high concentration of MMC resulted in a complete response and a median survival time comparable to that in recently published hepatic artery infusion studies with floxuridine and leucovorin. However, due to the hepatotoxicity we are continuing our studies with melphalan to further exploit the possible therapeutic benefit of ILP.