Patients treated with humeral head replacement (HHR) may require revision to total shoulder arthroplasty (TSA) due to glenoid arthrosis. This study characterizes the outcomes of revision TSA in patients who initially underwent HHR for nontraumatic glenohumeral arthritis.From 1982 to 2005, 68 shoulders underwent revision TSA for glenoid arthrosis. The initial HHR was performed for non-fracture-related arthritis. Revisions were grouped according to complexity for analysis. Stem revision and soft tissue reconstruction were assessed in relation to outcome.Pain scores decreased from 4.4 to 2.8, abduction increased from 85° to 116°, external rotation increased from 36° to 48°, and internal rotation remained unchanged. Survivorship free of repeat revision was 95.6%, 84.1%, and 72.9% at 1, 5, and 10 years. The Neer rating yielded 20 excellent, 10 satisfactory, and 38 unsatisfactory outcomes. No differences in motion, survivorship, or the Neer rating occurred between groups by procedure complexity. There was, however, less reduction in pain for the group requiring a humeral stem revision. Of the 9 shoulders with postoperative instability, 7 had compromised soft tissues preoperatively.Revision TSA after HHR results in decreased pain and increased motion. Result ratings, however, are quite variable and, in many cases, unsatisfactory. Stratification of the procedures according to complexity does not demonstrate differences in motion, satisfaction, or survivorship. Stem revision, however, results in reduced pain score improvement. Coexisting instability associated with subscapularis and anterior shoulder capsule damage may not be correctable using an unconstrained shoulder arthroplasty.