Locoregional therapies are useful in treating patients with colorectal cancer metastatic to the liver. A prospective randomized phase II trial of hepatic artery embolization versus hepatic artery chemoembolization was conducted to evaluate the response rates and toxicities of these therapies in the second-line setting. Patients were required to have biopsy-proven adenocarcinoma of the colon or rectum metastatic to the liver, with the liver as the sole or predominant site of metastatic disease. All patients had measurable disease and had failed at least one prior systemic chemotherapy treatment for metastatic disease. Patients were randomized to receive either embolization therapy with polyvinyl alcohol foam (Ivalon) administered as a single agent or chemoembolization using polyvinyl alcohol foam mixed with 750 mg/m2 of 5-fluorouracil and 9 million units of interferon. Drugs and embolic material were administered via the hepatic artery as a slurry with polyvinyl alcohol foam. Fifty eligible patients were enrolled. There were 24 patients in the chemoembolization arm and 26 in the embolization arm. Sixty-four percent of patients in both treatment arms had the liver as the sole metastatic site. The most common National Cancer Institute common toxicity criteria grade 3/4 toxicities were diarrhea (17%) and hepatic toxicity (8%). There was 1 (4%) treatment-related mortality due to a hepatic abscess. Four patients (15.4%) treated with embolization had a partial response (PR), and 5 patients (20.8%) treated with chemoembolization had a PR. The median survival for all patients was 11 months (95% confidence interval [CI], 8-15 months). Survival in patients with extrahepatic disease was 8 months (95% CI, 6-10 months). Survival in patients with liver-only metastases was 15 months (95% CI, 10-17 months). Embolization of the liver as second-line therapy in patients with liver-predominant metastases is safe and effective. Median survivals are comparable to other second-line therapies