Multiobjective optimizations of a micromixer with two-layer serpentine crossing channels were performed at two different Reynolds numbers by using a multiobjective genetic algorithm and surrogate modeling based on a Navier-Stokes analysis. The optimizations included three dimensionless design variables, namely, ratios of the diagonal channel width to the pitch length, the main channel width to the pitch length, and the channel depth to the pitch length. The design space was confirmed by a parametric study, and the Latin hypercube sampling technique was used to select design points within the design space. The Kriging meta-model was applied for surrogate modeling of the objective functions.Concave Pareto-optimal fronts representing the trade-offs between the objective functions were obtained through the optimizations.