Abstract The purpose of this study was to assess the clinical efficacy of high-dose levofloxacin plus rifampicin in the empirical treatment of non-tuberculous spondylodiscitis in an epidemiological context of low incidence of staphylococcal fluoroquinolone resistance. All consecutive adult patients with spondylodiscitis (January 2003 to December 2006) were empirically treated with high-dose levofloxacin (500 mg every 12 h normalised to renal function and optimised by means of therapeutic drug monitoring whenever feasible) plus rifampicin 600 mg every 24 h. Trough and peak plasma concentrations were targeted at 1–3 mg/L and 6–9 mg/L, respectively, to maximise the concentration-dependent activity of levofloxacin in bone. Follow-up was performed until 9 months after the end of therapy. Forty-eight patients were included. Eleven patients underwent a surgical approach for spine stabilisation. Among the 29 bacterial isolates, Staphylococcus aureus was the most frequent (65.5%) (all meticillin-susceptible strains). Tailored levofloxacin plasma exposure over time was ensured in most cases. Median treatment duration was 15.1 weeks. Overall response rates were: 77.1% at the intent-to-treat analysis; 84.1% among patients who completed therapy ( N = 44); and 96.3% among those receiving targeted therapy against documented levofloxacin-susceptible isolates ( N = 27). No patient had evidence of disease relapse at follow-up. Our findings suggest that high-dose levofloxacin regimens may be highly effective in the treatment of non-tuberculous spondylodiscitis and support its putative role in combination with rifampicin against S. aureus.