The purpose of this study was to compare the medium-term results for anatomic total shoulder arthroplasty by humeral component stem length. We hypothesize that the newer stemless implants may have comparable results to short-stem and conventional stemmed implants. The 12 most used anatomic total shoulder arthroplasty implants on the New Zealand Joint Registry were included in the study. Implants were categorized by stem length—conventional, short, and stemless. The primary outcome was revision up to 7 years postsurgery. Secondary outcomes included revision cause, implant survival, and early functional outcomes as evaluated by the Oxford Shoulder Score. Analysis was stratified by age and surgeon volume to control for potential confounding. A total of 3952 patients (conventional, 3114; short, 360; stemless, 478) were included in the study. No significant difference in revision rate per 100 component-years was found between stemless, short-stem, and conventional stemmed implants (revision rate per 100 component-years: conventional, 1.01 [95% confidence interval (CI) 0.89-1.14]; short, 0.54 [95% CI 0.25-1.03]; stemless, 0.99 [95% CI 0.51-1.74]). This finding was irrespective of patient age or surgeon volume. There were no cases of humeral loosening up to 7 years' follow-up and no cases of intraoperative humeral fracture in the stemless group. Functional outcomes at 6 months postsurgery suggested better outcomes in the stemless group compared with the conventional stem group (mean Oxford Shoulder Score: conventional, 39.4; stemless, 40.7; P value =.023). The medium-term survival of stemless implants for anatomic total shoulder arthroplasty appears comparable to short-stem and conventional stemmed implants. Further follow-up is required to understand the long-term survivorship and functional outcomes between these groups. [ABSTRACT FROM AUTHOR]