Accumulating experimental and clinical evidence has been obtained over recent years in support of the notion that the immune system has the potential to cure cancer. The most convincing example is the graft versus leukaemia effect observed after allogeneic haematopoietic stem cell transplantation. In the autologous setting, however, the isolation and expansion of naturally occurring tumour-specific T cells is a challenging task. Cancer antigens are often self-antigens and cancer-specific T cells are thus subject to selective mechanisms of central and peripheral tolerance. The significant advances in gene-transfer technologies developed over the last decade have offered new tools to overcome these limitations. Natural T cells can be genetically modified to generate high numbers of ‘supernatural’ tumour-reactive T cells from virtually every cancer patient. Supernatural T cells may express clonal receptors providing new specificities, factors increasing T-cell performance or safety factors enabling their elimination in case of toxicity. Technological improvements applied to novel concepts of T-cell biology and oncogenesis will allow to simultaneously equip T cells with different transgenes and expand a real ‘army’ of lymphocytes trained to selectively eradicate cancer cells.