This paper proposes an analysis and evaluation method for a human motion skill proficiency. In this method, the skill proficiency of individual human is analyzed and evaluated from its time series action force. The action force data of a skill is recorded at first. To record the data, master-slave robots on which bilateral control is implemented are utilized. By using the recorded data, skill proficiency is analyzed from two aspects. One is skill accuracy and the other is skill repeatability. Finally, the skill proficiency of individual is evaluated quantitatively by employing graph theory. For skill evaluation, a pocket billiard robotic system is constructed. The proposed method is applied to the evaluation of a pocket billiard stroke motion. The experimental results validate that the proposal abstracts one's skill characteristic, which indicates skill proficiency.