This retrospective study was conducted to analyze the local control and late complications in patients with squamous cell carcinoma of the maxillary sinus treated with postoperative radiation therapy following surgery. Between 1979 and 1998, 41 patients with squamous cell carcinoma of the maxillary sinus were treated with postoperative irradiation following partial or total maxillectomy. Tumor classification according to the TNM classification of the International Union Against Cancer (1997) was T2 in 6 patients, T3 in 21 patients, and T4 in 14 patients. Fourteen patients had negative surgical margins, 23 had microscopically positive margins, and 4 had grossly positive margins. Sixteen patients received preoperative intraarterial chemotherapy. The total dose to the primary tumor bed was 40-70 Gy (median: 54 Gy) with a fraction size of 2 Gy. The median follow-up time of the surviving patients was 93 months (range: 25-179 months). Local recurrence was observed in 17 patients (41%), and the 5-year actuarial overall survival and local control rates were 48% and 55%, respectively. In the univariate analysis, surgical margin status and total dose each had a statistically significant impact on local recurrence. For the patients with negative surgical margins, 8 of 9 (89%) patients achieved local control with a dose of 50-54 Gy, while 7 of 10 (70%) patients with microscopically positive margins achieved local control with a dose of 60-64 Gy. There were 11 late complications found in 9 patients; bone necrosis in 2, soft tissue necrosis in 2, trisumus: 2, cellulitis in 1, retinopathy in 1, and vision impairment in 3 patients. A total dose of 60 Gy or more was administered in all patients who suffered late complications except for 2 patients with vision impairment. These results indicated that an optimal dose of postoperative irradiation according to the surgical margin status was necessary to achieve local control for squamous cell carcinoma of the maxillary sinus following surgery. For patients with negative surgical margins, a total dose of 50-54 Gy in conventional fractionation was appropriate to achieve local control as well as to reduce late complications. On the other hand, a dose of 60 Gy or more was required for the patients with microscopic positive margins.