This study introduces a dynamic shoulder model, where forces were appl
ied to individual muscles in ten cadaveric specimens. The model provid
ed reproducible glenohumeral joint motion and thereby allowed the inve
stigation of active, glenohumeral joint mechanics. Forces were created
by servo-actuated hydrodynamic cylinders and applied to the deltoid m
uscle and to the rotator cuff through wire cables. Computerized regula
tion initiated precise, time controlled cycles of glenohumeral joint m
otion. The position of the glenohumeral joint in all spatial orientati
ons was measured and recorded using an ultrasonic sensor device. Repro
ducibility of glenohumeral joint motion was demonstrated on the basis
of five cycles of glenohumeral joint elevation. Repeatability variance
of position measurements for five cycles of elevation averaged 0.80 d
egrees for abduction, 0.75 degrees for anteflexion and 1.36 degrees fo
r internal rotation. Arm weight and force distribution at the shoulder
musculature were estimated according to the literature. In comparison
to estimated physiologic conditions, a one third increase of arm weig
ht led to a significant (p < 0.05) decrease of elevation of 20%, a one
third decrease of arm weight to an average increase of elevation of 1
8% (p < 0.05). Exclusion of the supraspinatus muscle caused a signific
ant (p < 0.05) 6% decrease of elevation of the glenohumeral joint. Wit
hout force applied to the subscapularis and infraspinatus;teres minor
muscles, elevation decreased 16% (p < 0.05). A decrease of glenohumera
l joint elevation of 25% resulted when force was applied to the deltoi
d muscle alone (p < 0.05).