BACKGROUND & AIMS: Aberrant DNA methylation is a common early event in neoplasia, but it is unclear how this relates to dysregulation of DNA (cytosine-5) methyltransferases (Dnmts). Here we use knock-in transgenic mice to investigate the consequences of intestinal epithelium-specific overexpression of de novo Dnmt3a. METHODS: A novel gene targeting strategy, based on the intestinal epithelium-specific, uniform expression of the A33 glycoprotein, is employed to restrict Dnmt3a overexpression in homozygous A33(Dnmt3a) mutant mice. RESULTS: A33(Dnmt3a) mice infrequently develop spontaneous intestinal polyps. However, when genetically challenged, tumor multiplicity in A33(Dnmt3a);Apc(Min) compound mice is 3-fold higher than in Apc(Min) mice. Although we observe a requirement for spontaneous loss of heterozygosity of the adenomatous polyposis coli (Apc) gene to trigger tumorigenesis in Apc(Min) mice, lesions in A33(Dnmt3a);Apc(Min) mice frequently retain the wild-type Apc allele. However, epithelia from normal mucosa and polyps of A33(Dnmt3a);Apc(Min) mice show hypermethylation-mediated transcriptional silencing of the Wnt antagonists Sfrp5, and to a lesser extent, Sfrp1 and increased nuclear beta-catenin alongside activation of the Wnt-target gene Axin2/Conductin. Conversely, enforced Sfrp5 expression suppresses canonical Wnt-signaling more effectively in wild-type than in Apc(Min) cells. CONCLUSIONS: Aberrant activation of the canonical Wnt pathway, either by mono-allelic Apc loss or transcriptional silencing of Sfrp5 is largely insufficient to promote polyposis, but epistatic interactions between these genetic and epigenetic events enables initiation and promotion of disease. This mechanism is likely to play a role in human colorectal cancer, because we also show that elevated DNMT3A expression coincides with repressed SFRP5 and enhanced AXIN2/CONDUCTIN expression in paired patient biopsies.