Background: The number of total shoulder arthroplasties has increased exponentially over the last ten years, creating a more prominent role for revision shoulder arthroplasty in the future. The main reasons for failure of shoulder arthroplasty can be classified as soft-tissue deficiencies, osseous deficiencies, component wear, or infection. We hypothesized that, despite appropriate surgical techniques, the outcome of revision total shoulder replacement can be predicted on the basis of the indication for the revision procedure. Methods: We conducted a retrospective review of seventy-eight shoulders that had undergone revision shoulder arthroplasty. The shoulders were divided into two categories: (1) those with osseous or component-related problems and (2) those with soft-tissue deficiency. Category 1 consisted of four cohorts of shoulders: twenty-two treated with revision of the glenoid component, sixteen treated with conversion of a hemiarthroplasty to a total shoulder arthroplasty because of glenoid arthrosis, eight treated with revision of the humeral stem, and four treated for a periprosthetic fracture. Category 2 consisted of five cohorts of shoulders: ten treated with rotator cuff repair following total shoulder replacement, four with a failed tuberosity reconstruction, four with cuff tear arthropathy, five with instability, and five with infection. Patients were evaluated with the UCLA subjective outcome instrument, the L'Insalata shoulder questionnaire, and a subjective satisfaction scale (maximum score of 5 points). Results: The average UCLA score was 21.4 points and the average L'Insalata score was 68.73 points for the seventy-eight shoulders that were analyzed. The average score on the subjective satisfaction questionnaire was 2.91 points. According to the UCLA scores, twenty-four revisions were considered to have had an excellent result; fifteen, a good result; twenty-four, a fair result; and fifteen, a poor result. The average scores for the category-1 shoulders were significantly better than those for the category-2 shoulders (p < 0.05). Of the different types of operations, revision or implantation of a glenoid component and open reduction and internal fixation of a periprosthetic fracture provided the best outcomes. Tuberosity reconstruction, hemiarthroplasty for treatment of cuff tear arthropathy, and revision due to infection had uniformly poor outcomes. Conclusions: In general, these results indicate that the outcome of revision shoulder arthroplasty can be predicted on the basis of the indication for the procedure. Component revisions, excluding humeral head revision for salvage, provide the best results, whereas soft-tissue reconstructions can be expected to yield poorer results overall. Level of Evidence: Prognostic Level II.