Genetically engineered mouse models (GEMMs) of human disease rarely take into account epigenetic alterations, which contribute to many human cancers. Now, Stephen Tapscott and colleagues compare genome-wide patterns of cancer-specific DNA methylation in human patients and three GEMMs of medulloblastoma (Epigenetics 8, 1254–1260, 2013). Using two independent methods to measure CpG methylation, they find, in contrast to the hypermethylation patterns previously observed in patients with medulloblastoma, that GEMMs had only modest increase in CpG methylation of gene promoters relative to wild-type controls. Whereas a human medulloblastoma tumor sample showed >60% increase in methylation at 121 loci, consistent with the authors' previous work, there were only 0–16 such loci in the GEMMs. They further extend these findings to mouse models of Burkitt lymphoma and breast cancer, which showed similar results. The authors confirmed, using immortalized mouse fibroblast cell lines, that the machinery necessary for promoter hypermethylation is present in mouse cells but is not activated in oncogene-driven GEMMs of cancer. The authors suggest that these data will be useful in developing more precise GEMMs for preclinical studies, which are especially crucial in identifying drugs that target the epigenome.