Capacitive deionization, CDI, is a new desalination technology which gained interest recently due to its economic advantages over existing technologies. To develop better electrode materials, this study focused on revealing CDI mechanisms and the most effective parameters. As current experimental techniques have limitations in producing well-defined pore arrangements, a multi-dimensional transient simulation tool was developed based on solving the Nernst-Planck equation with a convection term and Poisson's equation. The equations are solved using a coupled-implicit scheme, which does not require a very short time step due to the convergence problem. The present tool is verified through comparisons with theoretical solutions and other results. The key performance indicator (KPI) was defined as the salt removal rate under a fixed external electrode surface area, which relates to the capital cost of an electrode. The effects of the design parameters on the KPI are studied regarding the electrode gap distance, pore depth, pore diffusion coefficient, pore width, and flow velocity under stagnant water and flowing water conditions. It was found that the pore depth, pore diffusion coefficient and pore width were the most influential parameters on the KPI, while the gap distance and flow velocity had relatively small effects on the KPI. (C) 2011 Elsevier B.V. All rights reserved.