While it is clear that the skeleton is responsive to the demands imposed on it by habitual activity, the complexity of the response of bone to loading is not immediately apparent. Since the goal of adaptive changes in bone is the production of a structure that will resist loads with some adequate margin for safety, there will be different solutions in different bones. It is therefore important to consider what sort of loading is experienced by a bone physiologically, as the mass and architecture will depend on the combination of loading experience superimposed upon some genetically determined minimal baseline structure. We know for example that bones respond to high magnitudes of load more than low, high rates of strain change more than slow rates, and it is possible that bone is more sensitive to high frequencies of loading than low. Interestingly though after a relatively small number of load cycles, the response of bone to a given loading regimen becomes saturated so that larger numbers of cycles in the same bout do not increase the response. If however a saturating number of cycles is divided into several bouts in a single day, then that division causes a potentiated response which is greater than the same number in a single bout. This potentiation means that some evidence of earlier bouts of loading is retained by the skeleton which influences the effect of later bouts. The mechanism behind such a “memory” for exercise in bone is unknown. Some years ago, in an attempt to understand adaptive signalling mechanisms better, we used a subtractive technique (differential RNA display) and showed regulated expression of a component of a glutamatergic synapse in osteocytes a few hours after loading in vivo. This led us to explore expression of a range of molecules associated with glutamate signalling in the CNS, in bone cells, and we have shown that all cells in bone express such molecules to a greater or lesser degree. In vitro studies have shown that glutamate signalling is involved in both bone formation and resorption, but in vivo studies have been problematic, because of difficulties with drugs that cross the blood-brain barrier and affect behaviour. However recent studies using tissue specific gene knockouts have begun to yield data. One essential component of a glutamatergic synapse in the CNS is the NMDA type receptor – a glutamate gated channel. The NMDA receptor is a heteromer, comprising NMDA receptor 1 subunits and one of 4 NMDA receptor 2 subunits. The NMDA receptor 1 subunit (now known as Grin1 for glutamate receptor ionotropic – NMDA) is considered obligate and its deletion ablates NMDA receptor signalling in the CNS. By crossing a floxed Grin1 mouse with either osteocalcin or tartrate resistant acid phosphatase Cre strains, we have deleted Grin1 in osteoblasts and osteoclasts respectively. Surprisingly the phenotype of the OCCreGrin1flox mice is variable and mild, despite clear evidence for a role for glutamate in osteoblast differentiation in vitro. However, mice lacking Grin1 in osteoclasts and their precursors have evidence of tibial flare, and osteosclerosis, suggesting that absence of glutamate receptors in those cells affects their function significantly. Our current working hypothesis is that glutamate release by osteocytes and osteoblasts regulates bone resorption and contributes to the mechanism by which bone tunes its structure to activity.