In this work, a computational model of heat transfer and pressure drop is established for the design optimization of the annular involute-profile cross wavy primary surface (CWPS) recuperator in microturbine. The genetic algorithm method is employed to solve the optimization problem of annular CWPS recuperator with multiple design variables. Thus, an optimal design approach for this kind of recuperator is formulated. The validity of the computational model is checked by comparing the calculation results with the experimental data of C30 and C65 prototype recuperators (Capstone Turbine Corporation). With the total relative pressure loss as the optimization target, the developed method is applied to the optimal design of the recuperator in a 300 kW power-level microturbine. In comparison with the original design, the optimal design results in a significant reduction of the pressure loss in the case that all constraints are satisfied, which indicates the formulated optimal design approach is effective for the annular CWPS recuperator. By replacing the optimization target, the optimal design approach presented in this work can be also applied to the single-objective optimization for other targets (compactness, heat transfer area, weight, volume, effectiveness, etc.) and even the multi-objective optimization according to the particular requirements of different microturbines.