Aiming at improving the model accuracy for reliability assessment of protective devices, a new approach to estimate the parameters of the Weibull distribution model, integrating improved mean-rank method and the least square method, is proposed. It overcomes the shortcoming that the mean-rank method only considers the truncation-point position of each datum, and does not take into account the impact of the length of each failure interval on the analysis result. Based on pretreatment of failure data and censored data of protective devices, the shape and scale parameters of the Weibull distribution are estimated with the mean-rank method and the improved mean-rank method respectively, and the reliability function and the failure-rate function are obtained and analyzed. Taking the failure data and the censored data of a type of protective device as examples, parameter-estimation results of these two methods are compared, which verifies that the results obtained by the proposed method are more closely aligned with the practical operation status of the protective device. This work is supported by National Natural Science Foundation of China (No. 51877084), Natural Science Foundation of Hebei Province (No. E2018502063), and Fundamental Research Funds for the Central Universities (No. 2017MS096).