Epigenetic gene regulation is highly stable: epigenetic memory of gene expression states can persist over many cell generations and potentially for longer. However, epigenetic regulation is also flexible: genes that are subject to epigenetic regulation can respond dynamically to environmental and developmental signals. How can epigenetic regulation be both stable and flexible? I propose that the key lies in the highly dynamic nature of epigenetic systems. Over the last two decades it has become clear that the nucleus is an extraordinarily busy and noisy place: many proteins, including epigenetic regulators, are in constant motion, exchanging rapidly between chromatin bound and free states. Quantitative aspects of this motion are highly regulated. I propose that that to fully understand this regulation, epigenetics needs mathematics. We need ‘‘moving models’’ built of mathematical descriptions, which we can feed with measured values of quantities and mobilities of the components. A good model makes testable predictions that tell us whether our hypothesis makes sense. If it does not, we change the model. There has never been a better time to combine theoretical approaches with quantitative experiments. On the theoretical side, the last decade has seen a quiet revolution in the application of models built by physicists to the deep questions of epigenetics. On the experimental side, the advent of technologies that allow real time analysis at the single cell and single molecule level, together with those that enable targeted genome editing, allow precise perturbation and quantitative measurements at an unprecedented level. It is time for epigenetics to meet mathematics. I will give examples from work in the field and in my own lab, of how the fusion of experiment and theory has brought fresh insights into epigenetic regulation that go beyond intuition.