In order to alleviate the calculation catastrophe of bulk power system reliability evaluation， a coarse-grain parallel Monte Carlo simulation methodology and its asynchronous implementations combining convergence checking， based on HPC（high performance computation） Cluster and MPI（message passing interface） communication mechanism， are presented. Meanwhile， a web-based computation system using B/S 3-layer architecture and LAMP（Linux+Apache+MySQL+PHP）network environment is then proposed involving the assembling model of computation module on HPC Cluster for bulk power system reliability evaluation and the network database design for power system reliability information processing. The design and implementation for bulk power system reliability evaluation， in the shape of HPC and Web service， provide a promising approach to the power system analytical software based on grid computation.