Due to the low efficiency of book double-check in current book logistics, we propose a method of weight double-check. Using weight as a quantitative representation, this method considers book double-check as a classification problem. First, 3s principle and hypothesis testing are adopted to build the initial classifiers, based on the mean and variance of the weight of different classes. Then we construct the final classifier from the initial classifiers through Adaboost algorithm. This method overcomes the deficiency of initial classifiers, and improve the efficiency of current double-check approach.