A comparison of electromyographic (EMG) activity of muscles between and within subjects, and during separate occasions of testing, requires normalization. The most common way for generating the reference level used for normalizing shoulder EMG data is with a maximum isometric voluntary contraction (MVIC). The purpose of this study was to develop a parsimonious set of standardized tests that generate an MVIC in all the major muscle groups of the shoulder. Twelve muscles of the dominant shoulder of 15 subjects were examined using a combination of surface and intramuscular electrodes during 15 tests. The results indicated that many tests maximally activated more than one muscle simultaneously. Four tests were identified as being sufficient for generating an MVIC in the 12 muscles examined and are recommended as the standard set for normalizing shoulder muscle EMG: abduction 90 degrees with internal rotation ("empty can"), internal rotation in 90 degrees abduction ("internal rotation 90 degrees"), flexion at 125 degrees with scapula resistance ("flexion 125 degrees"), and horizontal adduction at 90 degrees flexion ("palm press"). The use of these shoulder normalization tests will make comparisons between shoulder EMG studies more reliable.