Abstract: In order to cope with the uncertainty which is introduced by massive distributed resources connecting to flexible distribution networks by layers, we proposed a probability scenario-driven distributed reactive power optimization method for flexible distribution networks. Firstly, we established a reactive power optimization model of flexible distribution networks with the goal of minimizing system losses. Secondly, considering the confidence constraints of 1-norm and ∞-norm, we constructed a distributionally robust reactive power optimization model of flexible distribution networks based on probability scenario ambiguity sets. On this basis, the distributed optimization model was used as the external framework for global coordination, and update iterative solution was solved by using a consistent acceleration gradient ADMM, while the distributionally robust optimization model for each sub region was used as the internal framework for solution using the CCG algorithm. The simulation results of an improved IEEE-33 bus system show that the proposed distributed reactive power optimization method for flexible distribution networks has good convergence, balancing economy, and robustness.