In this paper, locomotion of a biped robot operating in a physics-based virtual environment is evolved using a genetic algorithm, in which some of the morphological and control parameters of the system are under evolutionary control. It is shown that stable walking is achieved through coupled optimization of both the controller and the mass ratios and mass distributions of the biped. It was found that although the size of the search space is larger in the case of coupled evolution of morphology and control, these evolutionary runs outperform other runs in which only the biped controller is evolved. We argue that this performance increase is attributable to extradimensional bypasses, which can be visualized as adaptive ridges in the fitness landscape that connect otherwise separated, sub-optimal adaptive peaks. In a similar study, a different set of morphological parameters are included in the evolutionary process. In this case, no significant improvement is gained by coupled evolution. These results demonstrate that the inclusion of the correct set of morphological parameters improves the evolution of adaptive behaviour in simulated agents.