This paper presents a method to optimize domain shapes and thermal and flow boundary conditions on mixed convection heat transfer using genetic algorithm (GA) under distributed computing environment based on grid computing. GA increases the probability of finding the global optimum solution and avoids convergence to a local optimum which is a drawback of gradient-based methods. Furthermore, it is possible briefly to reduce computational time because of implicit parallelism of GA. GA method under distributed computing environment is applied to an example optimization problem. As a result, it is found that this method is able to obtain better solution than gradient-based method in less computational time.