Abstract: Based on improved particle swarm optimization (IPSO), an optimal method is proposed for needle roller bearings design so that the optimal dimensional parameters can be obtained when the bearings operates under complex conditions.By taking the operational conditions such as pollution, load, lubrication, speed, clearance, tilt and needle shape into account, a constraint optimal model is established which takes the modified rating life of needle roller bearings as the objective function.The random weighting strategy and the arccosine strategy are used to improve the inertia weight and adjust the learning factor, respectively.The constraints are solved by using the method of ensure particle legitimacy.As a result, the nonlinear optimization design of needle roller bearings was realized by IPSO.By analyzing the optimization results, the proposed method shows its effectiveness for needle roller bearings design.