To avoid the negative effects of structural parameters′ randomness on its properties, an optimization method of locking elastic sheet based on reliability sensitivity was proposed. Firstly, the finite element model and failure modes of the locking elastic sheet for a magnetic suspension control moment gyro were analyzed. Afterwards, considering the randomness of the structural sizes and material parameters of the elastic sheet, the reliability of the elastic sheet in the series failure mode were studied by using the response surface method and the first order second moment method. Furthermore, the calculating formulas of reliability sensitivity with respect to the mean and variance of random variables were derived. Eventually, the model of elastic sheet reliability optimization was built with the optimization strategy determined by the result of reliability sensitivity analysis. The difference influence of safety factor optimization and reliability optimization was analyzed. Experimental results indicate that the reliability optimization could get better optimization results compared with the safety factor optimization. The reliability sensitivity optimization method not only reduces the weight of the elastic sheet by 23.71% under the premise of ensuring the high reliability of the elastic sheet, but also reduces the weight of gyro room by 0.98 kg, which is beneficial to weight reduction of the magnetic suspension control moment gyro.