Seventy-four patients with liver metastasis from proved colorectal primary adenocarcinoma were entered into a prospective, randomized clinical trial to evaluate treatment with intra-arterial floxuridine compared with standard outpatient therapy with fluorouracil delivered by intravenous bolus injection. Eligible patients were randomized to hepatic arterial chemotherapy with an implanted infusion pump or systemic chemotherapy. No crossover between treatment arms was permitted, and patients were followed up to progression and death. Objective tumor response was observed in 48% of patients receiving intra-arterial floxuridine and in 21% of patients receiving intravenous fluorouracil. Time to hepatic progression was significantly longer in the group given intra-arterial therapy: 15.7 vs 6.0 months. However, time to overall progression (6.0 vs 5.0 months) and survival (12.6 vs 10.5 months) were not statistically different. Based on these data, we cannot recommend treatment with intra-arterial floxuridine as given in this study for metastatic colorectal cancer to the liver.