Abstract Background Pyogenic vertebral osteomyelitis (PVO) is rising in incidence, but optimal methods of investigation and duration of antibiotic therapy remain controversial. Methods We conducted a single-center retrospective cohort study of PVO at an Australian teaching hospital. We included all adults with a first episode of PVO between 2006 and 2015. PVO was defined based on the presence of prespecified clinical and radiological criteria. The main exposures of interest were investigation strategy and antibiotic treatment. The main outcome measures were duration of hospital admission, mortality during index admission, symptom resolution during index admission, and attributable readmission within 2 years. Results Of 129 included patients, 101 (78%) had a causative organism identified. Patients with an identified pathogen were more likely to be febrile (75% compared with 29%, P < .001) and had a higher mean admission C-reactive protein (207 vs 54, P < .001) compared with patients without an identified pathogen. However, they were less likely to experience an adverse outcome (death or attributable readmission within 2 years; adjusted odds ratio, 0.36; 95% confidence interval, 0.13–0.99; P = .04). Open biopsy of vertebral tissue had a higher diagnostic yield (70%) than fine needle aspirate (41%) or core biopsy (30%). Despite receiving a median of 6 weeks of intravenous antibiotics, only 15% of patients had a full recovery on discharge from index admission. Conclusions Clinical outcomes for patients with PVO were poor. Obtaining a microbiological diagnosis is associated with a better outcome. However, prospective and randomized studies are essential to establishing optimal investigation and treatment pathways.