Clinically effective cancer immunotherapy has been sought for more than 100 years and has been recently applied most successfully in strategies that passively deliver immune effectors such as monoclonal antibodies (anti-CD20 for lymphoma and anti-HER2/neu for breast cancer), donor lymphocyte infusions in chronic myelongenous leukemia and non-myeloablative allogeneic peripheral blood progenitor transplants for renal cell carcinoma. There is mounting enthusiasm for strategies employing active stimulation of antitumour immune responses. These include vaccines based on tumour antigen proteins and peptides, autologous, allogeneic or gene-modified tumour cells, dendritic cells and antigen-encoding viral vector constructs. Indeed, randomised Phase III clinical trials of autologous tumour cell vaccines for colorectal cancer demonstrated an improvement in disease free survival and a trend toward improved overall survival [1]. Despite these preliminary successes, it is clear that the many strategies under development cannot all be evaluated for survival benefit in large clinical trials that require many years, patients and resources to complete. This highlights the need to develop intermediate markers to help prioritise which agents to test in prospective randomised Phase III trials.