In view of the problems that the online judge is widely used and the existing paper-generating algorithm is low in efficiency, process complexity, and the quality of the paper-generating is not high. Study the artificial paper-generating, random function selection method, backtracking search method and genetic algorithm, and proposes a Hierarchical Solving paper-generating algorithm based on backtracking search. First, establish an algorithm model according to the idea of hierarchical solving pruning, and then the algorithm analysis model is set up with the goal is to improve efficiency and quality of the paper-generating, according to the type of questions, chapters and other indicators, the question bank is hierarchical solved to improve the efficiency of the paper-generating and the quality of the paper-generating. The experimental results show that with the increase of the amount of data, the hierarchical solving paper-generating algorithm based on backtracking search has obviously improved in the efficiency of paper-generating(repetition rate, coverage rate), the quality of paper-generating and the quality of paper(discrimination, reliability).