Introduction. Historically, a majority of prosthetic joint infections (PJIs) grew Gram-positive bacteria. While previous studies stratified PJI risk with specific organisms by patient comorbidities, we compared infection rates and microbiologic characteristics of PJIs by hospital setting: a dedicated orthopaedic hospital versus a general hospital serving multiple surgical specialties. Methods. A retrospective review of prospectively collected data on 11,842 consecutive primary hip and knee arthroplasty patients was performed. Arthroplasty cases performed between April 2006 and August 2008 at the general university hospital serving multiple surgical specialties were compared to cases at a single orthopaedic specialty hospital from September 2008 to August 2016. Results. The general university hospital PJI incidence rate was 1.43%, with 5.3% of infections from Gram-negative species. In comparison, at the dedicated orthopaedic hospital, the overall PJI incidence rate was substantially reduced to 0.75% over the 8-year timeframe. Comparing the final two years of practice at the general university facility to the most recent two years at the dedicated orthopaedics hospital, the PJI incidence was significantly reduced (1.43% vs 0.61%). Though the overall number of infections was reduced, there was a significantly higher proportion of Gram-negative infections over the 8-year timeframe at 25.3%. Conclusion. In transitioning from a multispecialty university hospital to a dedicated orthopaedic hospital, the PJI incidence has been significantly reduced despite a greater Gram-negative proportion (25.3% versus 5.3%). These results suggest a change in the microbiologic profile of PJI when transitioning to a dedicated orthopaedic facility and that greater Gram-negative antibiotic coverage could be considered.