Introduction: The GeneXpert® MRSA/SA SSTI (Methicillin Resistant Staphylococcus aureus/S. aureus skin and soft tissue infection) PCR test allows early detection of methicillin resistance in staphylococci. This test was developed for skin infections and has been evaluated for prosthetic joint infections but, to our knowledge, has not been evaluated for hardware infections outside of arthroplasties. Furthermore, we conducted a retrospective study in patients with non-prosthetic osteosynthesis hardware aiming: (1) to identify the diagnostic values of the PCR test compared to conventional cultures and the resulting rate of appropriate antibiotic therapy; (2) to identify the rate of false negative (FN) results; (3) to identify and compare the rates of failure of infectious treatment (FN versus others); (4) to search for risk factors for FN of the PCR test. Hypothesis: The PCR test allowed early and appropriate targeting of antibiotic therapy. Material and methods: The results of PCR tests and conventional cultures for osteoarticular infections of non-prosthetic hardware over four years (2012-2016) were compared to identify the diagnostic values of using the results of conventional culture as a reference and the rate of appropriate antibiotic therapies. Infectious management failures between the results of the FN group and the others were compared, and variables associated with a FN of the PCR test were identified. Results: The analysis of 419 PCR tests allowed us to establish a sensitivity of 42.86%, a specificity of 96.82%, a positive predictive value of 60% and a negative predictive value of 93.83%. Using the results of the PCR test for the targeting of postoperative antibiotic therapy, it was suitable for staphylococcal coverage in 90.94% (381/419). The rates of patients for whom infectious treatment failed were not significantly different between the FN group and the other patients (20.8% versus 17.7%, respectively; Hazard Ratio=1.12 (95%CI 0.47-2.69, p=0.79)). A skin opening during the initial trauma (p=0.005) and a polymicrobial infection were significantly associated with a risk of FN from the PCR test (p<0.001). Conclusion: The PCR test makes it possible to reduce the duration of empirical broad-spectrum antibiotic therapy during the treatment of an infection of osteosynthesis hardware but causes a lack of antibiotic coverage in 9.06% of cases. Level of evidence: III; diagnostic case control study.