Total shoulder arthroplasty (TSA) still suffers today from mid-term and long-term complications such as glenoid implant loosening, wear, humeral head subluxation/dislocation and implant fracture. Unlike the hip and knee joint replacements, the artificial shoulder joint has yet to offer a long-term satisfactory solution to shoulder replacement. With loosening being the number one reason for TSA revision, investigating methods of monitoring the glenoid implant loosening and investigate the effects of various design parameters on the loosening behaviour of the glenoid fixation is necessary to explore the problem. Several studies were carried out using in-vitro cyclic testing and FEA to; investigate failure progression and its correlation to quantitative measures in a 2D study (n = 60), investigating key glenoid design features in a 2D (n = 60) and 3D study (n = 20), investigating the validity of using bone substitute foam for studying glenoid fixation in a cadaveric study and investigating any correlation between failure and CT or in-vitro quantitative measures (n = 10). Visible failure was observed, for the first time, correlating to inferior rim displacement and vertical head displacement measures. CT failure was detected in 70% of specimens before visible failure was observed. Out of the design pairs tested; smooth-back/rough-back (range of roughnesses), peg/keel, curved-back/flat-back and conforming/non-conforming, roughening the back-surface to 3.4 μm or more improved fixation performance (p < 0.05). Roughening the back-surface changed the mode of failure from implant/cement failure inferiorly due to tensile/shear stresses, to cement/bone failure superiorly due to compressive/shear loading. Differences in the other design pairs were marked showing peg to perform better than keel, conforming over non-conforming and no difference in curved-back over flat-back, although these differences are marginal. Improvements in the standard testing method have also been suggested.