The reliability of tests for isometric strength of the shoulder joint in symptomatic subjects has yet to be established. For this purpose, interrater and intrarater agreement trials were undertaken to ascertain the reliability of manual muscle tests, a handheld dynamometer, and a spring-scale dynamometer for 5 different shoulder movements in symptomatic subjects. Intraclass correlation coefficients were calculated from a random-effects model. All movements tested with the handheld dynamometer demonstrated excellent reliability for the interrater trial (rho = 0.79-0.92). Excellent reliability was also demonstrated for elevation, external rotation, and internal rotation for the intrarater trial (rho = 0.79-0.96). For the interrater trial, measurement of the lift-off maneuver with the handheld dynamometer was significantly more reliable than with manual muscle tests (P =.002). In summary, the handheld dynamometer was the most reliable and discriminatory means for assessing strength of the rotator cuff in symptomatic subjects.