Background: In vitro wear studies demonstrate decreased polyethylene wear with a ceramic compared with a metal humeral head. In this in vivo study, we analyzed prospectively collected 5-year data, comparing the outcomes of 2 types of humeral implants. Methods: Sixty-two anatomic total shoulder arthroplasties (TSAs) (59 patients) were performed using the same glenoid component and 2 types of modular humeral implants (metal-head long-stem prosthesis in group A vs. ceramic-head stemless prosthesis in group B). Radiographic analysis for proximal humeral osteolysis and glenoid component radiolucent lines was performed at an average of 5.5 years. Functional outcomes were prospectively evaluated according to the visual analog scale score for pain; satisfaction level; American Shoulder and Elbow Surgeons score; active elevation; Shoulder Pain and Disability Index score; and Disabilities of the Arm, Shoulder and Hand score. Results: There were 39 TSAs in group A (mean age, 68.1 years) and 23 TSAs in group B (mean age, 67 years). Clinical improvement was similar for both groups. Some proximal humeral osteolysis occurred in 56% of all TSAs. Group A showed increased osteolysis compared with group B (72% vs. 30%, P ¼ .005). Glenoid radiolucent lines were far more frequent in group A, and group A showed a direct relationship between the extent of humeral osteolysis and the severity of glenoid radiolucent lines (P < .001). Conclusion: The stemless ceramic-head replacements showed fewer glenoid radiolucent lines and less humeral osteolysis than the long-stem metal-head replacements. In the group with metal-head replacements, correlation was shown between glenoid radiolucent lines and humeral osteolysis.