The optimal design of degradation experiments under the cost restraint is studied. The performance degradation process of products is defined with the degradation data distribution. Whether the total experiment cost exceeds the predetermined budget or not is set as the constraint condition, the objective is to minimize pth quantile variance of the degradation data, an optimal model for degradation experiment is established. A Monte Carlo simulation program is developed to obtain the optimal experiment scheme. An example of a metal crack degradation data is used to verify the efficiency of the method.