Using phosphogypsum lightweight coarse aggregate instead of natural gravel to prepare lightweight aggregate concrete is an effective technology to realize the comprehensive utilization of phosphogypsum resources. Based on the mix ratio design principle of lightweight aggregate concrete and BP neural network model, a method to predict the mechanical properties of small particle size phosphogypsum lightweight aggregate concrete was proposed. The results show that the compressive strength and splitting tensile strength of concrete decrease with the increase of net water cement ratio, increase with the increase of sand ratio, and decrease slightly with the increase of cement content. The order of significance of the three influencing factors is net water cement ratio, sand ratio and cement content. Appropriately reducing the net water cement ratio, using high sand ratio and low cement content can reduce the generation of pores and initial microcracks in the interfacial transition zone, and improve the overall mechanical strength. The constructed BP neural network model has a high accuracy in predicting the mechanical strength of small particle phosphogypsum lightweight aggregate concrete. The purpose of this study is to provide reference for mix ratio design optimization and mechanical strength prediction of phosphogypsum lightweight aggregate concrete.