BACKGROUND: In some cases, Helicobacter pylori infection persists even after three eradication treatments. AIM: To evaluate the efficacy of an empirical fourth-line rescue regimen with rifabutin in patients with three eradication failures. DESIGN: Multicentre, prospective study. PATIENTS: In whom the following three treatments had consecutively failed: first (PPI + clarithromycin + amoxicillin); second (PPI + bismuth + tetracycline + metronidazole); third (PPI + amoxicillin + levofloxacin). INTERVENTION: A fourth regimen with rifabutin (150 mg b.d.), amoxicillin (1 g b.d.) and a PPI (standard dose b.d.) was prescribed for 10 days. OUTCOME: Eradication was confirmed by (13) C-urea breath test 4-8 weeks after therapy. Compliance and tolerance: Compliance was determined through questioning and recovery of empty medication envelopes. Adverse effects were evaluated using a questionnaire. RESULTS: One-hundred patients (mean age 50 years, 39% men, 31% peptic ulcer/69% functional dyspepsia) were included. Eight patients did not take the medication correctly (in six cases due to adverse effects). Per-protocol and intention-to-treat eradication rates were 52% (95% CI = 41-63%) and 50% (40-60%). Adverse effects were reported in 30 (30%) patients: nausea/vomiting (13 patients), asthenia/anorexia (8), abdominal pain (7), diarrhoea (5), fever (4), metallic taste (4), myalgia (4), hypertransaminasemia (2), leucopenia (<1,500 neutrophils) (2), thrombopenia (<150,000 platelets) (2), headache (1) and aphthous stomatitis (1). Myelotoxicity resolved spontaneously in all cases. CONCLUSIONS: Even after three previous H. pylori eradication failures, an empirical fourth-line rescue treatment with rifabutin may be effective in approximately 50% of the cases. Therefore, rifabutin-based rescue therapy constitutes a valid strategy after multiple previous eradication failures with key antibiotics, such as clarithromycin, metronidazole, tetracycline and levofloxacin.