An investigation of the effect of tumor volume and total dose on local control following primary radiotherapy for nasopharyngeal carcinoma was carried out in order to estimate the radiation dose necessary to control a specific tumor volume. Between 1983 and 1996 a total of 104 patients underwent radiation therapy for nasopharyngeal carcinoma at the Department of Radiation Oncology of the University of Wuerzburg. Total doses of between 8 and 80 Gy (5 fractions per week) were administered. Complete CT-data on primary tumor size, total tumor dose (calculated by 3D- or quasi 3D-CT-based radiation planning computer) and on local control status in the follow-up period were available for 63 patients. Lymph node metastases were present in 38 of these patients and they were also entered into the study. Thus this study is based on a total of 101 tumor regions. A Poisson probability-based model was used for calculating the dose-response relationship. Assuming a correlation between tumor volume and the total dose necessary to obtain local control, the individual tumor volumes were rescaled to a 1 ml volume by introducing a volume-dependent modification factor for the applied dose, in order to eliminate the influence of different individual tumor volumes. All dose values given are based on a fractionation scheme of 2 Gy single dose, 5 fractions per week. Nineteen tumors and 11 lymph nodes were considered locally uncontrolled or recurrent. Without dose-volume modification, a weak dose-response correlation was found and a typical shallow dose-response curve was calculated with a 50% response dose (RD50) of 60.2 Gy and a normalized dose-response gradient (gamma50) of 3.2+/-0.62. After dose-volume modification and rescaling to a 1 ml tumor volume, a steep dose-response curve with an RD50 of 40.9 Gy and gamma50 of 8.2. was found. Tumor volume is a very important factor influencing local control in nasopharyngeal carcinoma. The rescaling procedure to a reference volume of 1 ml used in this study revealed a very steep dose-response relationship. This result suggests that the clinically observed smooth dose-response relationships may be explained by interindividual tumor volume heterogeneity. The additional dose necessary to control a tumor of the double volume is close to 5 Gy. With a total dose of 72 Gy (5x2 Gy/week), tumor volumes larger than 64 ml are unlikely to be controlled.