Abstract: To comprehensively enhance the aerodynamic and stealth performance of wide-speed-range variable-Mach-number waveriders, a multidisciplinary optimization design platform was constructed based on optimal Latin hypercube sampling, radial basis function surrogate models, and an improved multi-objective genetic algorithm. High-fidelity numerical simulations and a hybrid physical optics-method of moments approach were employed to evaluate the aerodynamic characteristics and radar cross-section, respectively. Sensitivity analysis indicated that the half-cone angle \delta has the most significant impact on the lift-to-drag ratio, with an increase leading to a decline in the ratio; the height distribution coefficient \phi exerts the strongest influence on electromagnetic scattering characteristics, with its increase capable of reducing the radar cross-section. Optimization results show that within the inflow Mach number range of 10 ~ 30, the optimized configuration achieves an improvement in lift-to-drag ratio of 4.03% ~ 4.72% compared to the baseline. Over a pitch angle range of ±50°, the average radar cross-section of the optimized configuration decreases across all angles, with the maximum reduction of 18.16% occurring at a pitch angle of 50°. At the typical 0° pitch angle, the average radar cross-section decreases from −32.51 dBsm to −34.30 dBsm, a reduction of 5.51%. Moreover, scattering peaks in key lateral regions (yaw angles of 30° ~ 60° and 120° ~ 150°) are significantly suppressed, leading to a systematic improvement in all-aspect stealth performance.