Background: Strengthening the supraspinatus is an important aspect of a rehabilitation program for subacromial impingement and tendinopathy. Many authors recommended empty-can (EC), full-can (FC), and prone full-can (PFC) exercises to strengthen the supraspinatus. However, no ultrasonography study has yet investigated supraspinatus muscle architecture (muscle thickness; MT, pennation angle; PA, fiber bundle length; FBL) in relation to supraspinatus strengthening exercises. Objects: The purpose of this study was to compare the architecture (MT, PA, and FBL) of the supraspinatus muscle during three different types of exercises (EC, FC, and PFC) using diagnostic ultrasound. Methods: Participants performed three different exercises: (A) EC; the arm was maintained at 60° abduction with full internal rotation in the sitting position, (B) FC; the arm was maintained at 60° abduction with full external rotation in the sitting position, and (C) PFC; the arm was maintained at 60° abduction with full external rotation in the prone position. Ultrasonography was used to measure the MT, PA and FBL of the supraspinatus. One-way repeated analysis of variance with Bonferroni’s post-hoc test was used to compare between the three exercises and the initial position of each exercise. Results: Compared with each initial position, the FC exercise showed the greatest mean difference in muscle architecture properties and the PFC exercise showed the least mean difference. Conclusion: The findings suggest that the FC exercise position may have an advantage in increasing the amount of contractile tissue or producing muscle power and the PFC exercise position may be useful in a rehabilitation program because it offers the advantage of maintaining the muscle architecture properties.