Data mining is used to analysis a large number of data. The study was using asset management of freeway flexible pavement and re-dug out new knowledge of pavement cracks by using decision tree of data mining. Road units were built per 20 meters per lane. Construction data (35,828 documents) was built and included road construction type, altitude and topography. Maintenance constructions data (36,439 documents) was built. The historical data included equivalent single axle loads, temperature and rainfall. Decision tree of C5.0 and C&R tree are different from the principle of calculation. The result of C5.0 was 92 rules and accuracy rate was 95%.The result of C&R tree was 21 rules and the accuracy rate was 80%. The results by C5.0 could be used to distinguish the reasons of pavement crack, and the results of C&R tree could be used to forecast the types of pavement crack.