Cryogenic liquid turbine expanders have been increasingly used in liquefied natural gas (LNG) production plants to save energy. However, high-pressure LNG commonly needs to be throttled to or near a two-phase state, which makes the LNG turbine expander more vulnerable to cavitation. Although some work has been reported on cryogenic turbomachine cavitation, no work has been reported on designing a cavitation-resistant two-phase LNG liquid turbine expander. Motivated by the urgent requirement for two-phase liquid turbine expanders, an effective design optimization method is developed that is well-suited for designing the cavitation-resistant two-phase liquid turbine expanders. A novel optimization objective function is constituted by characterizing the cavitating flow, in which the overall efficiency and local cavitation flow behavior are incorporated. The adaptive-Kriging surrogate model and cooperative coevolutionary algorithm (CCEA) are incorporated to solve the highly nonlinear design optimization problem globally and efficiently. The former maintains high-level prediction accuracy of the objective function but uses much reduced computational fluid dynamics (CFD) simulations while the later solves the complex optimization problem at a high convergence rate through decomposing them into some readily solved parallel subproblems. By means of the developed optimization method, the impeller and exducer blade geometries and their axial gap and circumferential indexing are fine-tuned. Consequently, cavitating flow in both the impeller and exducer of the two-phase LNG expander is effectively mitigated.