This article compares top-down and bottom-up methods for obtaining regional projections for policy analysis using computable general equilibrium models. In view of the computing and data problems of bottom-up, multisectoral models, a hybrid approach is proposed in which a partially regionalized computable general equilibrium model is used to drive a top-down regional equation system. The hybrid avoids some of the serious theoretical shortcomings of the top-down approach but is much less data demanding than a complete bottom-up model. An application, based on the ORANI computable general equilibrium model of Australia, is presented with comparative top-down and hybrid results.