Aims: The inertial measurement unit (IMU), as a sensor-based assessment tool, could provide objective and quantitative data for evaluating a patient with adhesive capsulitis (AC). The IMUs have advantages in simplification of implementation, cost, and computation complexity. We aimed to propose an IMU-based approach to extract statistical features for the assessment of AC in daily activity. Methods: Nine healthy subjects and nine AC patients participate in this experiment. The accelerometers are placed on the wrist and arm to measure the movement performance. Each subject is asked to perform three basic shoulder motions, including flexion, extension, and abduction. Eight types of features are extracted from the norm of accelerometer signals, including mean, standard deviation (SD), variation, maximum, minimum, range, kurtosis, and skewness. These features are explored to distinguish the differences in the movement performance between healthy subjects and AC patients. Statistical Analysis Used: Student's t-test and effect size (Cohen's d) are calculated to assess the reliability of the proposed evaluation approach. Results: The results show that the feature of SD extracted from the wrist can achieve significant differences and large effect sizes between healthy subjects and AC patients. Conclusion: We demonstrate the feasibility of the proposed IMU-based functional evaluation for the AC assessment using statistical features.