A multiobjective optimization of a micromixer with convergent-divergent sinusoidal walls has been conducted using flow and mixing analyses, surrogate modeling, and multiobjective genetic algorithm. The ratios of amplitude to wavelength of the sinusoidal walls, throat width to depth of the convergent-divergent sections, and diameter of the inner circular wall to wavelength were chosen as the design variables for optimization. The full-factorial method was used to discretize the design space. The mixing index and nondimensional pressure loss were selected as objective functions. Radial basis neural network functions were used to train the objective functions. The optimization was carried out at a Reynolds number of 30. A concave Pareto-optimal front representing the trade-off between the two objective functions was obtained. The analysis of representative designs along the Pareto-optimal front showed significant variation in the ratio of throat width to depth of the convergent-divergent sections, whereas the ratio of amplitude to wavelength of the sinusoidal walls maintained a nearly constant value. The concept of mixing effectiveness was used to select the most efficient designs considering both the mixing performance and pressure drop.