Abstract: We design and implement a parallel association rules mining method based on high-performance computing clusters (HPC clusters).The algorithm adopts the separation strategy to simply visit a local database only once,thus,the inter-processor communication I/O overhead is reduced.What's more,that using MPI/OpenMP hybrid programming mode can minimize data communication overhead on the network.By using ordinary PC structures,HPC cluster experimental results verify that the proposed algorithm based on HPC offers higher efficiency and has a good speedup.