This study proposed a new metric to assess the arm mobility based on the monitoring of daily activity using inertial sensors. Forty-one healthy participants and 21 patients suffering from rotator cuff tear disease were evaluated using a clinical questionnaire and one day measurement of the arm kinematics. Arm movements were detected from inertial data, and characterized by their velocity. Cumulative distribution of the velocities of each patient was compared to the reference distribution of controls, using the Kolmogorov-Smirnov distance. This distance was significantly correlated to clinical score. Moreover, it was used to compute the probability of "healthiness of arm movement" for each patient. The approach, based on the mobility of daily activity, offers an effective way to quantify the deviation of one patient from the group of controls.