Despite the steady and continuing growth of computing power and speed, the complexity and computational expense of engineering analysis codes maintains pace. Statistical techniques are becoming widely used in engineering design to construct approximations or metamodels of these analysis codes which are then used in lieu of the actual codes, facilitating optimization and concept exploration. Our purpose in this paper is to report results of ongoing research aimed at increasing the efficiency of computer-based engineering design through the use of spatial correlation metamodels to build global approximations of computationally expensive computer analyses. Three structural design examples are presented to test the predictive capability of these metamodels for use in design optimization. The reported results confirm that these spatial correlation metamodels can produce sufficient accuracy for optimization when used as global approximations.