In over 20 years since the discovery of altered methylation in cancer, many epigenetic alterations have been found in human cancer, including global and specific gene hypomethylation, hypermethylation, altered chromatin marks, and loss of genomic imprinting. Cancer epigenetics has been limited by questions of cause and effect, since epigenetic changes can arise secondary to the cancer process and its associated widespread changes in gene expression. Furthermore, mutations in the DNA methylation machinery have not been observed in tumors, whereas they have been for chromatin modification. To address the issue of human cancer etiology, we have taken a genetic approach to cancer epigenetics. One line of investigation has been on the disorder Beckwith-Wiedemann syndrome (BWS). We have found that loss of imprinting (LOI) of the autocrine growth factor gene IGF2 and of the untranslated antisense RNA LIT1, within the K(V)LQT1 gene, account for most cases of BWS, and that cancer risk is specifically associated with LOI of IGF2. Wilms' tumors, both in BWS and in the general population, involve LOI leading to an expansion of nephrogenic precursor cells. We have also developed an animal model for the role of LOI of IGF2 in cancer, showing that it cooperates with Apc mutations to increase cancer frequency, consistent with human data suggesting a severalfold increased cancer risk for this common epigenetic variant in the adult population. These data suggest that a major component of cancer risk involves epigenetic changes in normal cells that increase the probability of cancer after genetic mutation. They suggest a model of cancer prevention that involves the epigenetic analysis of normal cells for risk stratification and cancer prevention strategies.