The invention relates to a multi-agent game incremental distribution network source network load cooperative planning method, which comprises the following steps: respectively establishing planning decision models of a plurality of investment subjects; analyzing the static game behavior between DG investment operator and distribution network company according to the transfer relationship of threeinvestment agents, establishing a static game model when the game is in equilibrium; using robust optimization to deal with the uncertainty of DG output, and introducing the virtual game player Nature. According to the dynamic game behavior between the virtual game player and the distribution network company, the game reaches the equilibrium state, and the dynamic game model is established on thisbasis. Considering the multi-agent game of DG investment operators, distribution network companies, power users and Nature, a dynamic-static joint game layout for incremental distribution network planning is formed and a dynamic-static joint game model is built on the basis. The iterative search method is used to solve the Nash equilibrium point, and the programming decision model based on game theory is solved to get the final planning scheme.