OBJECTIVE Vertebral osteomyelitis (VO) is a compelling clinical entity for clinicians because of its insidious and indolent course, which makes diagnosis difficult. METHODS All patients with a suspected diagnosis of VO were analyzed over an 8-year period (January 2009 to January 2017). The UDIPROVE protocol (UDIne PROtocol on VErtebral osteomyelitis) was applied in all cases. The primary endpoint was the performance of the UDIPROVE protocol to obtain the causal bacteria of infection. RESULTS During the study period, 133 episodes of confirmed VO were observed. The etiology of infection was obtained in 73.6% of cases: 70.5% were gram-positive, 16.3% were gram-negative, and 13.2% were mycobacteria. 18F-Fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) showed that for tubercular VO, the median standard uptake value (SUV) was higher when compared with VO caused by other bacteria. Clinical cure at the end of therapy was reported in 85.7% of patients. Previous antimicrobial therapy and a delay of more than 5 days in performing biopsy were associated with an undiagnosed etiology of VO. Targeted antibacterial therapy and follow-up with FDG-PET/CT were associated with clinical cure at the end of therapy, while the involvement of more than two vertebrae and inadequate drainage were associated with failure. CONCLUSIONS Rigorous application of the UDIPROVE protocol allowed the causative pathogens of VO to be obtained - at about twice the rate reported in the literature. The use of FDG-PET/CT for the follow-up of infection was more reliable when compared to magnetic resonance imaging (MRI).