Based on data envelope analysis (DEA),the labor productivity was decomposed into 4 acpects:technical efficiency,pure technical progress,changes of scale efficiency of capital per labor and capital deepening.The rule extraction based on decision tree classification algorithms was used to get the effects of labor productivity’s affecting factors and to study how to affect provincial industrial labor productivity.The research finds that the most important factor affecting provincial industrial labor productivity is the capital deepening.The second factor is the technical efficiency,and the next factor is the scale efficiency of capital per labor.Pure technical progress almost doesn’t affect the difference between provincial industrial labor productivities.