A multi-phase, multi-component, thermal and transient model is applied to simulate the operation of a passive direct methanol fuel cell and optimize the design. The model takes into consideration the thermal effects and the variation of methanol concentration at the feeding reservoir above the fuel cell. Polarization and constant current cases are numerically simulated and compared with experiments for liquid feed concentration, membrane thickness, water management and air management systems. Parameters considered when determining an optimal design include power density, fuel utilization and energy efficiencies and water balance coefficients. An optimal liquid feed concentration is determined to be 2.0 mol kg(-1). which achieved a maximum power density of 21 mW cm(-2) and a fuel utilization efficiency of 63.0%. An optimal design of a cell uses a thick membrane (Nafion 117) to reduce methanol crossover and two additional cathode GDLs to improve the water balance coefficient and efficiency of the cell. This combination results in a power density of 23.8 mW cm(-2) and a water balance coefficient of -1.71. An air filter may also be added to improve the efficiency and water balance coefficient of the cell, however, a small loss in power density will also occur. Using an Oil Sorbents air filter the water balance coefficient is increased to -0.85, the fuel utilization efficiency is improved by 27.35% and the maximum power density decreased to 21.6 mW cm(-2). (C) 2009 Elsevier Ltd. All rights reserved.