For the home position calibration of a 4-DOF parallel robot (Cross-IV robot), a fast calibration approach based on the rotary angle errors of end-effector is proposed. Based on the closed-loop vector equations of a single limb, a mapping model between the complete home position errors and the errors of end-effector is established. By the decomposition of the error transfer matrix, a fast home position errors identification model of the robot is constructed, which only needs to measure the rotary angle of end-effector errors by the rotary encoder. To further maximize the measuring efficiency and improve the robustness of the identification matrix, an optimized selection scheme of the measuring points is proposed. Detailed simulation verifies the robustness and correctness of the proposed method. The verification experiment based on a laser tracker shows that the position error of end-effector is decreased to 1.312 mm and the rotary angle error is decreased to 0.202° after calibration, so the home position calibration method is simple and effective.