The prognostic value of age, sex, symptoms, histopathology, lymphocyte count, skin tests, immunocompetence (lymphocyte response profile to a spectrum of PHA concentrations), and stage was evaluated in 35 previously untreated patients with Hodgkin's disease by multiple linear regression and logistic analysis. Immunocompetence exhibited the highest correlation (R = 0.537) with survival status and was required as a common denominator for deriving best sets of two or more variables. In addition, immunocompetence contributed to all other variables combined (p = 0.023), whereas stage did not (p = 0.116). Immunocompetence, age, symptoms, and histopathology generated a highly discriminant (R = 0.784) model not improved by stage or by other variables (p greater than or equal to 0.159). The utility and generalizability of this model are shown by a correct classification of 91.2% of cases according to expected versus actual survival status and by a predicted correlation (R) of 0.71, respectively, neither improved by sex, lymphocyte count, skin tests, or stage. In comparison, the conventional triad of stage, symptoms, and histopathology correctly classified only 70.6% of cases and showed actual and predicted correlations with survival status of R = 0.550 and R = 0.51. We conclude that immunocompetence is a powerful discriminant risk factor in Hodgkin's disease that exerts a pivotal role on survival and serves as a basis for models of greater discriminant power and generalizability than the conventional stage-based evaluation triad. Immunocompetence-based models are expected to provide a more discriminating basis for clinical evaluation, prediction of prognosis, and treatment selection for patients with Hodgkin's disease.