Reports using the Constant-Murley score often lack details regarding how strength measurement was performed, which may have an important impact on the results obtained. A description of the measurement is especially important when patients are unable to adopt the prescribed position of 90 degrees shoulder abduction, leading to invalid or highly discrepant scores. We investigated the influence of shoulder abduction and torso position on strength measurement with special emphasis on intraobserver reliability. Thirty-three healthy volunteers were examined using a handheld dynamometer (Isobex) in 30 degrees , 60 degrees , and 90 degrees abduction with or without stabilization of the torso. A linear correlation between strength measurements in different degrees of abduction was observed, with values at 90 degrees abduction on average 15% lower than at 60 degrees and 45% lower than at 30 degrees . In multivariate analysis, gender and weight had a substantial impact on the strength of individual measurement positions. Intraobserver reliability was related to arm and torso position (intraclass correlation coefficients, 0.71-0.93) and was highest for 90 degrees abduction without torso stabilization. These findings emphasize the importance of standardized torso and arm positions ensuring high reliability when performing the strength measurement of the Constant-Murley score.