Previous analyses of findings from the Gastrointestinal Tumor Study Group (GITSG) study GI 6175, a four-arm randomized trial of patients with adjuvant colon cancer, have revealed no efficacy for chemotherapy or immunotherapy, but Dukes' Stage was found to have prognostic significance in both survival and disease-free survival. In this study, the exophytic tumor as primary type, maximal tumor dimension, and degree of differentiation were examined for prognostic importance. In a model simultaneously including treatment effects, significance levels for prognosis in total survival were: Dukes' stage (P less than 0.0001), exophytic tumors (P = 0.05), maximal dimension (P = 0.005), and well versus poor differentiated tumors (P = 0.06). This observation about tumor length, adjusted for stage and morphology, provides evidence that size is prognostic in colonic tumors. In addition, an interaction between tumor morphology and immunotherapy, adjusted for Dukes' stage, was noted. In 100 patients with exophytic tumors, significant survival and disease-free survival benefit (P less than 0.02) from immunotherapy occurred. Tumor morphology is found to be an important prognostic variable and should be carefully considered in the construction of future trials in this disease site.