The data mining technology enables the user to extract interested knowledge form high-magnitude data. Now most structural data are stored in multiple data table in the relational database. In this paper, the authors propose a multi-relational association rule mining algorithm based on user guidance to extend the traditional association rule method. With ID propagation idea, multiple tables can be directly associated for analysis without physical connection. The concept of user guidance is introduced to improve the mining efficiency and precision. This algorithm can support the relational database. The running time is far less than it of the ILP-based multi-relational association rule algorithm.