Despite recent improvements in the treatment of metastatic colorectal cancer, few patients are cured and the response rates to second-line treatments are poor. Onyx-015, an oncolytic virus, was administered to patients with metastatic colorectal cancer by hepatic artery infusion. No dose-limiting toxicities were observed in the phase I/II studies. Onyx-015 can kill tumor cells by mechanisms that are distinct from chemotherapeutic agents and may therefore have activity among patients who have failed first-line chemotherapy. The 24 patients included in this analysis had failed first-line therapy with 5-FU/leucovorin, 79% of the patients failed two or more regimens and 58% had failed treatment with Irinotecan. Despite the extensive prior therapy, the median survival of these patients was 10.7 months, 46% were alive at 1 year and two patients (8%) had partial responses. In all, 11 patients (46%) had stable disease at the completion of the four planned viral treatments (3 months). The median survival of this group of patients was 19 months, suggesting that stable disease may be an important predictor of benefit with oncolytic viruses. Eight of the 11 patients with stable disease at 3 months demonstrated a unique radiographic pattern of transient enlargement of tumor masses (10-48%) after the initial infusions of Onyx-015, followed by radiographic evidence of extensive tumor necrosis and regression. The initial enlargement and subsequent tumor necrosis resulted in a prolonged time to achieve objective tumor regression. In addition, the transient enlargement of the tumor masses may have resulted in premature removal of responding patients. Treatment of eight patients was stopped prior to completion of the planned four treatments due to presumed progression as defined by standard radiographic criteria (>25% increase in tumor size). Functional imaging, such as positron emission tomography (PET) scans, may help distinguish clinical responses from progressive disease following treatment with oncolytic viruses. Onyx-015 may benefit patients with refractory colorectal cancer and additional studies that include PET scans to assess clinical response are warranted.