An understanding of the mechanisms that explain the initiation and early evolution of colorectal cancer should facilitate the development of new approaches to effective prevention and intervention. This review highlights deficiencies in the current model for colorectal neoplasia in which APC mutation is placed at the point of initiation. Other genes implicated in the regulation of apoptosis and DNA repair may underlie the early development of colorectal cancer. Inactivation of these genes may occur not by mutation or loss but through silencing mediated by methylation of the gene's promoter region. hMLH1 and MGMT are examples of DNA repair genes that are silenced by methylation. Loss of expression of hMLH1 and MGMT protein has been demonstrated immunohistochemically in serrated polyps. Multiple lines of evidence point to a "serrated" pathway of neoplasia that is driven by inhibition of apoptosis and the subsequent inactivation of DNA repair genes by promoter methylation. The earliest lesions in this pathway are aberrant crypt foci (ACF). These may develop into hyperplastic polyps or transform while still of microscopic size into admixed polyps, serrated adenomas, or traditional adenomas. Cancers developing from these lesions may show high- or low-level microsatellite instability (MSI-H and MSI-L, respectively) or may be microsatellite stable (MSS). The suggested clinical model for this alternative pathway is the condition hyperplastic polyposis. If colorectal cancer is a heterogeneous disease comprising discrete subsets that evolve through different pathways, it is evident that these subsets will need to be studied individually in the future.