To overcome the excessive communication and calculation problems in the bulk power system reliability evaluation with the static task allocation technique， the rules of reliability index square coefficients varying with the sampling times in Monte-Carlo method are analyzed. Based on the analyzing results， a parallel computation model for reliability evaluation of bulk power system is proposed using a dynamic task allocation technique， which is designed using the De Moivre-Laplace central limit theorem and the curve fitting method. Using the simulating information in the parallel task allocation process， a conservative evaluation technique for total sampling times can be used to maximize allocated processes in a single task and minimize the total communication times. Case studies show that the proposed dynamic task allocation method can improve the efficiency of parallel computation.