We performed a retrospective study of 106 patients with carcinoma of the soft palate who were treated at two university hospitals. Computer analysis using a new interactive data base program called MING was made to determine Berkson-Gage survival and Gehan tests of statistical significance. Statistically significant associations with an increased survival included the following: smaller lesions, a clinically negative neck examination. well- and moderately well-differentiated histopathologic features, radiation therapy dose of less than or equal to 6,300 rads, absence of a simultaneous primary, and surgical salvage. No statistically significant differences were seen with age, sex, stage, or the number of days during which patients were treated with radiation therapy. There appears to be a need for a prospective, multi-institutional, randomized therapy study to solidify treatment policy. Consideration should be given to combine surgery-radiation vs radiation therapy alone.