Despite the spectacular contributions to knowledge made by molecular biology during the last half century, cancer research has not delivered an agreed explanation of how malignant tumours originate. The models assiduously investigated in molecular terms largely reflect waves of fashion, and time has revealed their inadequacy: cancer is (1) not caused by the direct action of oncogenes, (2) not fully explained by the impairment of tumour suppressor genes, (3) not set in motion by mutations controlling the cell cycle, (4) not governed by the dependence of malignant tumours on an adequate blood supply and (5) not triggered by a failure of programmed cell death. But there is now strong evidence that cancers may have their origin in mutations that block the execution of critical steps in the process of normal differentiation. Cancer, thus seen, is not initially a disease of cell multiplication, but a disease of differentiation. The evidence for this point of view should now be explored. BioEssays 27:833–838, 2005. © 2005 Wiley Periodicals, Inc.