Résumé : The aim of this multicentric prospective randomized clinical trial was to study the efficacy of hepatic artery ligation (HAL) with and without portal infusion (PI) of 5-FU in patients with liver metastasis of colorectal origin. Seventy-four patients were randomized. Sixty-seven were fully evaluable. Thirty-five patients were eligible in the HAL + PI of 5-FU group and 32 in the HAL alone group. The 5-FU infusion had to be discontinued for technical reasons in 13 patients. Complications of HAL were relatively high, including four hepatic failures (WHO grading greater than 2). Side effects of chemotherapy were limited. Five patients out of 30 had a partial response (WHO criteria) and one patient had a complete response in the group treated by HAL and PI of 5-FU. Only one patient had a partial response in the HAL alone group. Median survival for both groups was 12 months. Median time to progression for both groups was 6 months. This study did not show any advantage of delivery using the portal route in addition to hepatic artery ligation in terms of progression nor in survival of patients.