Aiming at the disadvantage of assessing service life of relay protection devices with the same yardstick, an assessment method of optimum service life is proposed. At first, device defects are classified into early defect, accidental defect and aging defect according to the defect cause and the failure rate characteristic of devices. Field data is used as random censoring test data to solve occurrence times expectation of device defect. The expectation of early defect is fixed value, accidental defect accords with exponential distribution and aging defect accords with Weibull distribution. Then, the function of unit time cost is built considering the effect of condition based maintenance policy and configuration to assess optimum service life. At last, the feasibility is proved utilizing the field data of transmission line protection devices. This work is supported by Natural Science Foundation of Fujian Province (No. 2016J01219).