In this paper, a design methodology combining coupling matrix representation of filters, neural models and space-mapping techniques is presented for further enhancement of optimization efficency of microwave filters. Neural models are developed for both initial dimension generation and design parameter sensitivity analysis. Combining neural models of filter substructures with space-mapping optimization, the total number of EM simulations of the complete filter structure is significantly reduced. The improvement in efficiency over conventional method is demonstrated using simulation and measurement results of both end-coupled and side-coupled waveguide dual-mode pseudo-elliptic filters. The total CPU times for design and optimization are reduced by 50% to 70 %. V C 2011 Wiley Periodicals, Inc. Int J RF and Microwave CAE 22:159-166, 2012.