The randomness of distributed wind generator（DWG）and the uncertainty of the load are firstly modeled.Based on that，a chance constrained programming method is proposed in the selection of site location and capacity in the distribution system aiming to minimize the total annual cost.The probabilistic power flow based on the Latin hypercube sampling-based Monte Carlo simulation（LHS-MCS）is applied to verify whether the planning scheme satisfies the chance constraints on the node voltage operation range and the feeder current limit.A procedure of solving the planning model by shuffled frog leaping algorithm（SFLA）is designed.The example of a33-bus distribution network validates the feasibility of the model，the effectiveness of LHS-MCS and the high efficiency of SFLA.