Preclinical models may be relatively uninformative about the magnitude of effect of an immunologic intervention in human cancer, as well as about the optimal dose and regimen. In human cancer there may be limitations as to the amount of tumor that can be used as starting material. The composition of the tumor specimen may vary within each tumor and between individuals with the same tumor type. Patients may vary in their ability to respond to the vaccine. Finally, the optimal route of administration and the dose, schedule and duration of administration must be defined. If in early human trials using this approach the anti-tumor effect is not sizeable and clearly evident, an adaptive program should be designed to determine the optimal parameters for the vaccine’s preparation and administration and to identify biomarkers for those individuals likely to benefit.