Cancer immunotherapy strategies often fail because of immunosuppressive mechanisms present in the tumour-bearing host. Adoptive T-cell transfer therapy circumvents this problem by activating tumour-specific CD8+ T cells in vitro and transferring them back into the patient. Classically, effector T cells have been used in these studies because of their potent anti-tumour activity. However, it is becoming apparent that highly activated effector cells may become terminally differentiated, display impaired proliferation and survival in vivo, and mediate short-term anti-tumour effects. In contrast to effector cells, memory cells have enhanced proliferative potential and survival, and the potential to provide more robust and enduring protection against tumours. Here, we discuss key studies in the field of adoptive T-cell transfer, along with some of our own results relating to this area. Based on the body of existing research, it is clear that CD8+ T cells with memory potential are superior to terminally differentiated effectors in mediating successful tumour clearance. Opinions remain divided as to whether the central memory or effector memory T-cell subset is capable of providing the best protection against tumours. We propose that as these cell types have different but complementary benefits for the anti-tumour immune response, the ideal cell population to use for adoptive T-cell transfer should consist of a heterogeneous mixture of memory cells.