It is emerging that the environment, particularly in early life, can cause long-lasting epigenetic changes that are accompanied by latent effects on health outcomes. Furthermore, the reversibility of some epigenetic marks in animal models indicates that, once recognised in early life, such epigenetic changes may be reversible, providing potential avenues for disease prognosis, prevention and treatment. Twin studies, which were originally used to calculate gross components of genetic and environmental influence on phenotype, are now being used to associate specific genetic and epigenetic variants with specific human conditions and diseases. Epigenome-wide association studies of phenotypically discordant, genetically ‘identical' monozygotic twins have the power to focus solely on epigenetic association with disease and are providing information about gene-environment interaction and variable penetrance. Going beyond association, such studies can help generate epigenetic biomarkers to predict disease long before clinical onset, which has important implications for disease prevention. Furthermore, when epigenetic information on a genome scale is combined with other ‘omics', twin studies will be a strong force for the improvement of human health.