Staphylococcus aureus is responsible for life-threatening conditions, while in the meantime it has rapidly acquired resistance to several antibiotic classes. In the context of an effective empirical antibiotic therapy, an accurate evaluation of the resistance rates of S. aureus may be critical. The aim of this study was to determine the resistance rates of S. aureus in the years 2015-2018 and to assess the impact of specimen stratification on the resistance rates. We have retrospectively analysed S. aureus strains isolated from blood, bronchial aspirate, pus, sputum and urine collected from hospitalized and ambulatory care patients. The comparison between resistance rates from 2015 to 2018 and among different specimens was assessed by Fisher's exact test followed by Benjamini and Hochberg's correction of the p-values. Higher resistance rates were detected for penicillin followed by oxacillin, levofloxacin, erythromycin and clindamycin. Differences in the annual resistance rates were not statistically significant after the BH's correction. The comparison between cumulative S. aureus resistance rates stratified by specimens showed some statistically relevant differences among the five specimen types. In particular, p-values were statistically significant for clindamycin, erythromycin, gentamicin, levofloxacin, oxacillin, penicillin and vancomycin. Annual resistance rates of S. aureus clinical isolates remained constant over the course of time. Moreover, the stratification of the data by specimen may significantly impact on the evaluation of the resistance rates, at least for some antibiotics. Therefore, if the number of data is high, stratification by specimens may be recommendable to better approach an empirical antibiotic therapy.