Colorectal cancer still represents one of the most common causes of morbidity and mortality among Western populations. The adenomatous polyposis coli (APC) gene, originally identified as the gene responsible for familial adenomatous polyposis (FAP), an inherited predisposition to multiple colorectal tumors, is now considered as the true "gatekeeper" of colonic epithelial proliferation. It is mutated in the vast majority of sporadic colorectal tumors, and inactivation of both APC alleles occurs at early stages of tumor development in man and mouse. The study of FAP has also led to one of the most consistent genotype-phenotype correlations in hereditary cancer. However, great phenotypic variability is still observed not only among carriers of the identical APC mutation from unrelated families but also from within the same kindred. The generation of several mouse models carrying specific Apc mutations on the same inbred genetic background has confirmed the genotype-phenotype correlations initially established among FAP patients, as well as provided important insights into the mechanisms of colorectal tumor formation. Here we review the major features of the available animal models for FAP and attempt the formulation of a hypothetical model for APC-driven tumorigenesis based on the observed genetic and phenotypic variability in mouse and man.