The forward kinematics solution for problem of parallel robot mechanism is the basis of application of robot mechanism, and it is one of the difficulties in the study of kinematics. Aimed at the numerical method and analytical method of the complex and solving difficulty and sometimes the uncertainty in solution, it was put forward the method of multilayer feed forward neural network for solving the forward kinematics of three degree of freedom (3DOF) 5P4R parallel mechanism. The inverse kinematics solution was regard as the trained sample, and the vector variable in study of BP algorithm was used, and then the nonlinear mapping from driver workspace to the moving platform variable space was achieved, so the parallel mechanism forward kinematics solution was obtained. At last, a set of simulation instance is given. It can be seen that the validity and feasibility of this method.