This study was aimed at the performance of micro methanol steam reformer for channel widths optimization by simplified conjugate gradient method (SCGM) with minimum objective function of velocity standard deviation. A threedimensional numerical model and optimal algorithm named simplified conjugate gradient method (SCGM) were built for predicting the dominating factors in channel widths. In the optimization searching process, the simulation model only includes the continuity and the momentum equations for decreasing the computational time. When the optimum channel width of different inlet flow rate and its distributions are obtained by SCGM with minimum objective function, the species and energy equations are applied to evaluate the performance of steam reformer. Distributions of velocity were predicted, and the methanol conversion ratio was also evaluated. In addition, the mole fraction of CO contained in the reformed gas, which is essential to prevent the catalyst layers of fuel cells from poisoning, is also investigated. Furthermore, the present simulation model was also compared with existing experiment data, and the close agreement was found. Results show that the methanol conversion ratio can be improved from 54.2% to 62.6% for the particular case at 0.6 cc/min flow rate if the channel widths are optimized. On the other hand, the mole fraction of carbon monoxide is increased from 2.85 % (original) to 5.21 % (optimal) at Q&=0.6 cc min-1.