In order to improve the position precision of moving platform end effector of linear driven parallel robot, based on structure and kinematic model of parallel mechanism, the dimension errors, driving errors and assembly errors of parallel robot which influenced accuracy of the moving platform end effector were analyzed, the non-linear mapping from the actuated variables in joint space to the pose of the end-effector in operating space was established with analytic method. On the basis of the sensitivity error model, an error source selection scheme was proposed according to the global sensitivity evaluation index, which was defined to evaluate the error source on the position error of the moving platform in given workspace, and the main error sources that affected the location accuracy was selected. It was found that the position error of the moving platform was basically the same between all error sources and selected error sources by randomly simulating the dimension errors, driving errors and assembly errors of the parallel robot with Monte Carlo method, which verified the consistency of the evaluation index. By taking laser tracker as measurement tool, the position precision of the moving platform end effector of the parallel mechanism was improved remarkably after detecting the selected main errors of the parallel robot and calibrating the forward kinematic model，which verified the validity and feasibility of the error source selection scheme and reduced the complexity and computation of the error parameter identification. It had greatly important guidance to the error compensation of the complex structure.