Frost has suggested that functional loading controls bone mass and form as a thermostat controls temperature--by homeostatic regulation (i.e., by negative feedback). Yet, the literature contains many results that appear incompatible with the "mechanostat" hypothesis. We propose that a different type of regulation--epigenetic--is important in many aspects of bone adaptation. Epigenetic regulation, as we describe it, involves positive feedback loops and promotes differentiation as it forces elements of a system to choose between two extreme levels called attractors. Our review of bone adaptation data suggests that lamellar bone formation is regulated homeostatically, whereas the formation of woven-fibered bone or fracture callus is regulated epigenetically, that is, woven bone formation is brought about by a positive feedback loop that stimulates osteoblasts to a state of greater individual activity. This positive feedback loop may involve transforming growth factor beta (TGF beta), for which autocrine induction has been demonstrated in vivo, as well as other factors, including insulin-like growth factor (IGF) and prostaglandin E2 (PGE2). The importance of this model is that it provides a mechanism for many unexplained nonlinear events that have been observed in bone adaptation experiments. Furthermore, it provides insights into the genesis of woven bone, which is a critical step in the process of bone healing and regeneration.