Since the 1990s, topology optimization has been used to computationally investigate the bone remodeling under the assumption that bone remodeling progresses such that the bone material is used in a structurally optimal way. Therefore most concerns in applying topology optimization to bone remodeling include the difficulty of handling large-scale problems and associated huge computational cost. In this paper, we applied a recently developed topology optimization algorithm, design space optimization (DSO), to bone remodeling simulation in order to determine trabecular architecture in human proximal femur with higher computational efficiency. We represented the full trabecular architecture in human proximal femur using a two dimensional micro-FE model with 50 mu m pixel resolution and performed simulation under three load cases in daily activities. From the quantitative comparison with conventional topology optimization results as well as the actual trabecular architecture, it was shown that DSO produced structurally equivalent trabecular architecture with shorter computing time and smaller memory requirement. As future research, a three-dimensional bone remodeling simulation will require the preparation and management of tens of millions of FE elements, and therefore DSO would be essential to handle such a massive problem as a "more efficient" algorithm. (C) 2009 Elsevier B.V. All rights reserved.