Since the cloning of the first tumor suppressor gene 22 years ago, we have learned a great deal about the role of tumor suppressor pathways in human cancer. One general principle is that some tumor suppressor pathways (e.g., p53 and Rb pathways) are inactivated in virtually every human cancer. Thus, one might predict that inheritance of a genetic lesion in such a pathway would cause the rapid onset of tumors originating from different tissues. However, this is not true for the Rb pathway. Children with a defective copy of the RB1 gene show increased susceptibility to retinoblastoma but not to other developmental tumors of the nervous system. Moreover, after RB1 inactivation, certain retinal cell types are more susceptible to tumorigenesis than others. Our recent studies on the role of the Rb family of genes in retinal development and retinoblastoma have led to a new hypothesis that explains this paradox. We propose that cells that require the Rb family for their cell fate specification and/or differentiation are less susceptible to tumorigenesis than those that do not require the Rb family for these processes. If correct, this hypothesis would allow us to predict which cell types in the developing nervous system are susceptible to tumorigenesis after inactivation of the Rb family and may establish a general principle of tissue- and cell type-specific susceptibility to tumorigenesis. In this perspective, we discuss our recent findings that have changed our views on tumor initiation and progression following Rb family inactivation.