Local tumor control is analyzed in a series of 181 patients treated with definitive megavoltage radiotherapy (RT) for histologically proven squamous cell carcinoma of the oropharynx. Considerable variation in treatment time stemmed from the general use of a split-course technique in 49 patients treated from 1978 to 1985. Incomplete follow-up, in those patients alive and well at the termination of the study or who have died from metastases or intercurrent disease before developing a local recurrence, was allowed for by using a multi-variate mixture model. The tumor control probability (TCP) after radiotherapy showed a significant dependence on the following tumor and treatment characteristics: (a) tumor size: the number of tumor target cells increases approximately as the fourth root of estimated tumor volume; (b) sex: the estimated TCP in males is lower than in females with the same characteristics; (c) histopathological differentiation: well-differentiated tumors have a lower TCP than poorly and intermediately differentiated; (d) hemoglobin concentration: patients in the upper normal range have a significantly higher TCP than others; (e) total dose: there is a significant dose-response relationship; and (f) overall treatment time: TCP decreased with increasing overall time, the dose equivalent of proliferation with 2 Gy per fraction was 0.68 Gy/day with 95% confidence limits [0.05, 1.3] Gy/day. The TCP did not depend significantly on subsite within the oropharynx or nodal disease at presentation. The data were consistent with an alpha/beta ratio of the linear-quadratic model of 10 Gy.