Abstract: In order to explore a simulation-based quantitative analysis method for the waterfront public spaces, this research builds the systematic analysis framework of “model construction - problem diagnosis - prediction optimization” based on the multi-agent behavior simulation technology. Taking the transformation of the Zhiyang segment, North Bund, Shanghai as an example, the research firstly determines the key elements and attractiveness weights influencing behavior activities through preference analysis, establishes the parameter relationship between simulated space and agent particle, and builds a multi-agent model based on necessary operation and fitting. Then the research carries out the itemized scenario simulation, diagnoses public space problems, and puts forward optimization and improvement measures. After that, the research improves single spatial elements, and compares and selects element combinations, based on which proposes the optimal recommendation for different orientations. Finally, the research selects similar combinations and compares their status before and after transformation, to verify the feasibility of the method developed. The research shows that the quantitative analysis method based on multi-agent behavior simulation is effective for waterfront public spaces. Having the behavior simulation technology innovatively extended and applied to the research of recreation behavior in waterfront public spaces, the aforesaid method can achieve the scientific prediction of the future use status of waterfront public spaces through simulation.