Understanding the surface stability and crystal growth morphology of hydroxyapatite is important to comprehend bone growth and repair processes and to engineer protein adsorption, cellular adhesion and biomineralization on calcium phosphate based bone grafts and implant coatings. It has generally been assumed from electronic structure calculations that the most stable hydroxyapatite surface is the (001) surface, terminated just above hydroxyl ions perpendicular to the (001) crystal plane. However, this is inconsistent with the known preferential growth direction of hydroxyapatite crystals and previous experimental work which indicates that, contrary to currently accepted theoretical predictions, it is actually the (010) surface that is preferentially exposed. The surface structure of the (010) face is still debated and needs reconciliation. In this work, we use a large set of density functional theory calculations to model the interaction of water with hydroxyapatite surfaces and probe the surface stability and resistance to hydrolytic remodeling of a range of surface faces including the (001) surface and the phosphate-exposed, calcium-exposed, and hydroxyl-exposed terminations of the (010) surface. For the (001) surface and the phosphate-exposed (010) surface, dissociative water adsorption is favorable. In contrast, the hydroxyl-terminated (010) surface will not split water and only molecular adsorption of water is possible. Our calculations show, overall, that the hydroxyl-terminated (010) surface is the most stable and thus should be the predominant form of the hydroxyapatite surface exposed in experiments. This finding reconciles discrepancies between the currently proposed surface terminations of hydroxyapatite and the experimentally observed crystal growth direction and surface stability, which may aid efforts to accelerate biomineralization and better control bone-repair processes on hydroxyapatite surfaces. (C) 2014 Elsevier B.V. All rights reserved.