Polymer electrolyte fuel cells, which have been developed for mobile equipment, are mainly direct methanol fuel cells (DMFC). From the standpoint of safety to a human body, it is desirable to use ethanol as fuel. In order to improve the performance of direct ethanol fuel cells (DEFC), it is necessary to investigate the fuel flow pattern in the anode flow channel with many sharp turn corners in the fuel cell, and to improve the channel shape. So, in this study, the velocity distribution of the water flow in a micro-channel was numerically analyzed using the authors' GTT code and the calculated results were verified by the velocity measurement with a line LDV system. For reducing the pressure loss in consideration of the suppression of flow stagnation in the channel, it was attempted to optimize the sharp turn corner shape by three-dimensional numerical flow analysis with a genetic algorithm (GA) and a shape with less pressure loss was found.