In the case of no-fault data,the failure probability combined with the Bayesian theory to study the confidence limit analysis to correct the unilateral confidence reliability evaluation. Based on the Weibull distribution,the distribution curve method was adopted to analyze,and the hierarchical priors distribution was constructed to give the comprehensive E-Bayes estimation of the failure probability,and the estimated value of distribution parameters were obtained by combining the least square method. According to the statistical property of the parameters,the pivot quantity of the confidence interval is constructed,and the interval estimation and reliability confidence lower limit of the Weibull distribution characteristic parameters are given.Taking the classical bearing test as an example,this method was compared with the calculation results of other methods,which proves the rationality and feasibility of this method in evaluating confidence reliability.