Abstract
I present a new hypothesis for cancer progression, based on observations with experimental tumors and on our growing understanding of the regulation of gene expression in mammalian cells. The experimental observations demonstrate that progression has a stochastic course and is associated with profound perturbations of cell differentiation. The hypothesis proposes that an initial event (such as the activation of an oncogene) alters the state of the regulatory network that controls the expression of cellular genes, directing it to evolve in a direction not consonant with the developmental program of the genome. Possible consequences of this epigenetic hypothesis for cancer research are discussed.