In most applications of multiple distributed control problems, higher precise relative states are easier to be obtained compared to absolute states. Based on this, a new relative states based distributed receding horizon formation control method was designed. The proposed distributed control algorithm was implemented by online solving a nonlinear optimal control problem through predicting the relative states trajectory. Compared to the traditional distributed receding horizon control algorithm, the most attracting advantages of the proposed algorithm are 1) the reduced computational burden and 2) the control precision improvement due to introducing less measurement errors. Also, under some conditions, the convergence can be ensured for most of multiple distributed systems with nonlinear models. At last, the algorithm is used in multiple robot formation control system in Matlab simulation environment to verify the method.