An artificial fish swarm algorithm based on Pareto multi-objective optimization is proposed for optimal sensor placement in structural health monitoring. The modal independence，damage sensitivity and damage redundancy are firstly utilized to establish the sensor multi-objective optimization function. Then the rear-end and foraging behaviors in the artificial fish swarm algorithm are improved，and the external file sets are introduced for the centralized processing of the mutually non-dominating solutions in the optimization process. The Pareto solution with the closest Euclidean distance at the ideal point is considered as the final optimal solution. A planar frame structure is finally used as a numerical study to verify the proposed artificial fish swarm algorithm based on Pareto multi-objective optimization for sensor optimal placement. The results obtained from the proposed method give a fairly uniform spacing for the sensor locations，and the information obtained by the measurements is more comprehensive，with low redundancy and good mode independence. The damage detection results also indicate the robustness of the proposed method.