In view of the slow convergence and low accuracy of the artificial fish-swarm algorithm forthe optimal design of truss structures, the Logistic equation was used to initialize the fish group to im-prove the efficiency and quality of solution on the initial stage of the algorithm, and the inertia weight adjustment strategy of particle swarm optimization was used to adjust the artificial fish's vision and to step to raise the speed and accuracy of optimization in the process of operation. Then the improved ar-tificial fish-swarm algorithm was applied to the truss structural optimization，and the optimal design model was established with the section size of the truss as the design variable，the minimum structural weight of structure as the objective function. The Matlab was used for the optimization analysis of the model, and the result was compared with those of other algorithms. It is showed that the algorithm is improved in the convergence rate and searching precision and the effect is obvious especially on the initial stage.