Objective In order to develop quantitative rehabilitation evaluation system for upper limb motor function based on the current end-effector-based rehabilitation robot training system, this study develops quantitative evaluation parameters for upper limb motor function based on the circle-drawing point-to-point movement tasks and validates the effectiveness of thoseevaluation parameters through the human-robot interactive experiments regarding the different movement directions. Methods Based on the existing hardware and software system of rehabilitation robot, this research utilizes interaction software to develop two types of human-robot interaction tasks including circle-drawing and point-to-point movements through the users' controlling on the end-effector. Eight volunteers participate in the experiments. Through the acquisition and analysis on motor's motion data, evaluation parameters are designed for quantitative evaluation on upper limb motor function. Statistical t-tests are applied to compare the differences between different circle-drawing groups ( clockwise and counterclockwise) and different point-to-point groups ( outer-lines group and declining-lines group) of evaluation parameters in order to validate their effectiveness.Results In the circle-drawing experiments on the eight volunteers, the result of roundness scores for counterclockwise group ( 91. 64±3. 45) is significantly higher than those of clockwise group ( 87. 84±3. 98) . In the point-to-point experiments on the seven volunteers, the results of difference in the total trajectory length and maximum value of normal deviation are significantly lower in the outer-lines group than in the declining-lines group. The results of average velocity and average acceleration are significantly higher in the outer-lines group than in the declining-lines group. The average value of normal deviation, precise positioning ability at end point and completion time are not significantly different between two different point-to-point movement groups with different directions. Conclusions The roundness score designed in this study could effectively evaluate the motor function of health adults ' upper limbs during the circle-drawing movements by controlling the rehabilitation robot. The difference in the total trajectory length, maximum value of normal deviation, average velocity and average acceleration could effectively evaluate the motor function during the point-to-point movements. By comparing with the previous study, the motor controlling function of human upper limbs could be different when the movements are in the different body-relative position. This might be affected by the upper limb's functional position.