Background Replicating a total shoulder arthroplasty in laboratory is a difficult task due to complex geometry of the structures and degrees of freedom of the joint. Implanted joint shoulders have been investigated using numerical tools, but models developed lack of experimental validation. The objective of this study was to develop a finite element model that replicated correctly an experimental simulator of an implanted joint shoulder based on the comparison of measured and calculated strains. The methods used include a non-cemented Anatomical Comprehensive© Total Shoulder System that was implanted in 4th generation composite bones. The finite element model designed replicates adequately the experimental model. Both models included the most important muscles of shoulder abduction and the same boundary conditions (loads, fixation, and interface conditions). Strain gauge rosettes were used to measure strain responses on the shoulder in 90° abduction. The results of linear regression analysis between numerical and experimental results present a high correlation coefficient of 0.945 and a root-mean-square-error of 35 µε, suggesting adequate agreement between the experimental and the numerical models. Small strains were obtained and changes in load distribution from posterior to anterior region were observed. As conclusion we can say that the experiments allowed good replication of the finite element model, and the use of strain gauges is suitable for numerical-experimental validation of bone joints.