Purpose: This Phase II trial evaluated the safety and efficacy of neoadjuvant chemotherapy (NAC) with S-1 and oxaliplatin (SOX) plus bevacizumab (Bev) in patients with colorectal liver metastasis (CRLM). Methods: Patients with initially resectable CRLM received four cycles of SOX plus Bev as NAC. We adopted the R0 resection rate as the primary endpoint, and the threshold R0 resection rate was set at 80%. Results: Between December 2010 and August 2014, 61 patients were enrolled in this study and all started NAC. The completion rate of NAC was 82.0%. Three patients (4.9%) developed severe liver dysfunction caused by NAC and one patient finally decided against resection. Three patients (4.9%) were judged as having progressive disease during or after NAC and did not undergo liver resection. Among 57 patients who underwent liver resection after NAC, three patients were diagnosed with CRLM by pre-treatment imaging modalities and received NAC although a final pathological diagnosis was another malignant disease or benign condition. Finally, 47 of the 54 patients (87.0%) with resected CRLM achieved R0 resection. The pathological complete response rate of the 54 patients was 13.0%, and 31.5% were judged as pathological responders. However, the R0 resection rate of 77.0% in the entire cohort did not meet the endpoint. Conclusions: NAC with SOX plus Bev has an acceptable toxicity profile and achieved a satisfactory pathological response. However, accuracy of pre-operative diagnoses and liver dysfunction caused by NAC were serious problems. Easy introduction of NAC for initially resectable CRLM should not be performed.