We address the challenging problem of optimizing building design for daylighting performance to minimize lighting loads, which is one of the two largest sources of energy consumption in buildings, besides thermal loads. We present a methodology to optimize complex building design for daylighting performance that uses dynamic climate-based lighting simulations. We compare four optimization algorithms as implemented in GenOpt: Simplex Algorithm of Nelder and Mead with the Extension of O'Neill (SA), Hooke Jeeves (HJ), Particle Swarm Optimization using Inertia Weight (PSOIW), and a hybrid PSO Constriction/Hooke Jeeves (PSOC/HJ) algorithm. The results indicate that the design space region of optimum performance is typically found quickly by all algorithms, but converge is sometimes slow. SA and HJ typically reached convergence in fewer function evaluations than PSOIW and PSOC/HJ; however, they did not consistently find solutions close to the overall best solution found. PSOIW found the best overall solution and consistently found other solutions close to it, but always took longest to converge. PSOC/HJ reached convergence more quickly than PSOIW and consistently found solutions slightly below the overall best solution found. While our work presented here focused on optimizing for daylighting performance in isolation of thermal envelope loads, the best daylighting design had annual summed lighting and thermal loads 33% lower than the best design optimized for thermal performance without consideration for daylighting. (C) 2015 Elsevier B.V. All rights reserved.