PURPOSE: We performed superselective intra-arterial infusion chemotherapy (SIC) according to a protocol in which drug distribution is evaluated by the use of interventional radiology (IVR)-computed tomography (CT) system, and the chemotherapy is combined with medium-dose conformal radiation therapy (CRT). We analyzed retrospectively the factors that affect the midterm survival ratio, including local response, for stage III and IV squamous cell carcinomas of the oral cavity. MATERIALS AND METHODS: Forty consecutive patients with stage III and IV squamous cell carcinomas of the oral cavity and who had undergone both SIC and CRT were enrolled. A microcatheter was placed in the appropriate feeding artery of the tumor and cisplatin (50mg/body) was infused twice. CRT was administered with a dual-energy (4 and 10 MV) linear accelerator. The total and daily doses delivered were 30 and 2.0 Gy, respectively. Histopathologic effects were classified into five grades: grade 0 or 1 was defined as a poor response, and grade II or higher as a good response. Age, sex, stage, local response to treatment, mode of invasion and lymph node metastasis were analyzed, and differences in the midterm survival ratio were assessed. RESULTS: The 3-year survival ratio of the 40 cases was 67%. A good local response (III or IV) was achieved in 75% of the cases. The survival ratio of the good local response group was significantly better than that of the poor response group (p=0.04). Mode of invasion (p=0.03) and lymph node metastasis (p=0.01) were also predictive of survival. In the multivariable analysis of survival, however, no variables including good local response (p=0.12), were predictive. CONCLUSION: Our new protocol improved local response, but it did not contribute to the survival ratio.