Monte Carlo simulation method combined with simple random sampling is easy to use; and higher accuracy can be obtained only with a large enough sample size. Consequently， in order to achieve lowers error， the computational speed will have to be reduced and the computational costs remain very high. An effective sampling method， Latin hypercube sampling， is integrated into the probabilistic load flow calculation to increase the sampling efficiency of Monte Carlo simulation by improving the sample values coverage of random variables input spaces. The effectiveness and efficiency of the proposed method is proven by the comparative tests in the IEEE 14-bus system and IEEE 118-bus system. Compared with simple random sampling， it needs a much smaller sampling number to get a specified accuracy， and can effectively evaluate the statistical parameters and probability distribution of output stochastic variables. At the same time， the advantages of Monte Carlo simulation are preserved， and computational cost is dramatically reduced when dealing the stochastic problems of the same stochastic variables.