Background: To examine the rates and predictors of deep periprosthetic infections after shoulder hemiarthroplasty. Methods: We used prospectively collected institutional registry data on all primary shoulder hemiarthroplasty patients from 1976-2008. We estimated survival free of deep periprosthetic infections using Kaplan-Meier survival curves. Using univariate Cox regression analyses, we examined the association of patient-related factors (age, sex, body mass index), comorbidity (Deyo-Charlson index), American Society of Anesthesiologists grade, underlying diagnosis, and implant fixation with the risk of infection. Results: A total of 1,349 patients, with a mean age of 63 years (SD, 16 years), 63% of whom were women, underwent 1,431 primary shoulder hemiarthroplasties. Mean follow-up was 8 years (SD, 7 years). Fourteen deep periprosthetic infections occurred during the follow-up, confirmed by medical record review. The most common organisms were Staphylococcus aureus, coagulase-negative Staphylococcus, and Propionibacterium acnes, each accounting for 3 cases (21% each). The 5-, 10-, and 20-year prosthetic infection-free rates were 98.9% (95% confidence interval [CI], 98.3%-99.5%), 98.7% (95% CI, 98.1%-99.4%), and 98.7% (95% CI, 98.1%-99.4%), respectively. None of the factors evaluated were significantly associated with risk of prosthetic infection after primary shoulder hemiarthroplasty, except that an underlying diagnosis of trauma was associated with a significantly higher hazard ratio of 3.18 (95% CI, 1.06-9.56) for infection compared with all other diagnoses (P = .04). A higher body mass index showed a non-statistically significant trend toward an association with higher hazard (P = .13). Conclusion: The periprosthetic infection rate after shoulder hemiarthroplasty was low, estimated at 1.3% at 20-year follow-up. An underlying diagnosis of trauma was associated with a higher risk of periprosthetic infection. These patients should be observed closely for development of infection. © 2012.