Abstract: To address strong constraint coupling and difficult feasible-waypoint search in cooperative multi-UAV path planning in complex three-dimensional environments, as well as directional convergence during exploitation, fixed exploration–exploitation allocation, and insufficient local search in the starfish optimization algorithm (SFOA), a multi-strategy improved starfish optimization algorithm (IMSFOA) is proposed. The success-history based adaptive differential evolution (SHADE) parameter-memory mechanism and current-to-pbest/1 guidance are introduced to diversify exploitation directions? sliding-window success-rate feedback dynamically allocates search resources? and covariance-guided local search improves accuracy around promising solutions. IMSFOA is compared with seven algorithms and evaluated by ablation experiments on 30-dimensional CEC 2017 functions, and is then applied to three-UAV cooperative three-dimensional waypoint planning considering terrain, obstacles, inter-UAV safety distance, and arrival synchronization. Results show that IMSFOA achieves the lowest average comprehensive cost and an 83.33% success rate in the standard scenario. Wilcoxon rank-sum tests indicate significant overall advantages over SFOA and six other representative algorithms, demonstrating good optimization accuracy and stability.