Recent in vitro studies reported the exceptional ability of some bacterial species to form biofilm-like aggregates in human and animal synovial fluids (SF), but evidences from infected clinical samples are still lacking. In this study, we investigated whether this bacterial phenotype was present in infected SFs collected from joint infections and if it was maintained in in vitro settings. SFs sent for culture to the Laboratory of Microbiology of our institute were directly analyzed by means of confocal laser scanning microscopy (CLSM), and the infective agents were isolated for further in vitro tests. Moreover, sterile SF was collected from patients who did not receive previous antibiotic therapy to investigate the formation of bacterial aggregates, together with biofilm and matrix production on a titanium surface. Finally, antibiotic susceptibility studies were performed by using bovine SF. Four Staphylococcus aureus, one Staphylococcus lugdunensis, and one Prevotella bivia strain were identified in the infected SFs. The CLSM analysis showed that all staphylococci were present as a mixture of single cells and bacterial clumps surrounded by an exopolymeric substance, which comprised SF-derived fibrin, while all P. bivia cells appeared separated. Despite that, differences in the ability to aggregate between S. aureus and S. lugdunensis were observed in clinical SFs. These different phenotypes were further confirmed by in vitro growth, even though the application of such ex vivo approach lead all staphylococci to form exceptionally large microbial aggregates, which are several folds bigger than those observed in clinical samples. Planktonic aggregates challenged for antibiotic susceptibility revealed a sharp increase of recalcitrance to the treatments. Although this is still at a preliminary stage, the present work confirmed the ability of staphylococci to form free-floating biofilm-like aggregates in infected SF from patients with joint infections. Furthermore, the obtained results pointed out that future in vitro research on joint infections will benefit from the use of human-or animal-derived SF. Even though this approach should be carefully validated in further studies comprising a larger microbial population, these findings pose new challenges in the treatment of infected native and prosthetic joints and for the approach to new investigations.