Changes in joint architecture and muscle loading resulting from total shoulder arthroplasty (TSA) and reverse total shoulder arthroplasty (RSA) are known to influence joint stability and prosthesis survivorship. This study aimed to measure changes in muscle moment arms, muscle lines of action, as well as muscle and joint loading following TSA and RSA using a metal-backed uncemented modular shoulder prosthesis. Eight cadaveric upper extremities were assessed using a customized testing rig. Abduction, flexion, and axial rotation muscle moment arms were quantified using the tendon-excursion method, and muscle line-of-force directions evaluated radiographically pre-operatively, and after TSA and revision RSA. Specimen-specific musculoskeletal models were used to estimate muscle and joint loading pre-and post-operatively. TSA lateralized the glenohumeral joint center by 4.3 ± 3.2 mm, resulting in small but significant increases in middle deltoid force (2.0%BW) and joint compression during flexion (2.1% BW) (p < 0.05). Revision RSA significantly increased the moment arms of the major abductors, flexors, adductors, and extensors, and reduced their peak forces (p < 0.05). The superior inclination of the deltoid significantly increased while the inferior inclination of the rotator cuff muscles decreased (p < 0.05). TSA using an uncemented metal-backed modular shoulder prosthesis effectively restores native joint function; however, lateralization of the glenoid component should be minimized intra-operatively to mitigate increased glenohumeral joint loading and polyethylene liner contact stresses. Revision RSA reduces muscle forces required during shoulder function but produces greater superior joint shear force and less joint compression. The findings may help to guide component selection and placement to mitigate joint instability after arthroplasty.