To compare the efficacy and safety of bismuth-containing quadruple therapy with tetracycline or amoxicillin for rescue treatment of Helicobacter pylori.The study was a non-inferiority trial of H. pylori eradication with at least two previous treatment failures. Subjects were randomized to receive 14-day therapy with b.i.d. lansoprazole 30 mg and bismuth 220 mg, plus metronidazole 400 mg q.i.d and amoxicillin 1 g t.i.d (amoxicillin group) or tetracycline 500 mg q.i.d (tetracycline group). Antimicrobial susceptibility was assessed by the agar-dilution method. Primary outcome was H. pylori eradication at 6 weeks after treatment.In all, 312 subjects were randomized, 13 were lost to follow-up; 29 violated the protocol. The intention-to-treat, per-protocol, and modified intention-to-treat eradication rates were (amoxicillin) 88.5% (138/156, 95% confidence interval (CI) 83.4-93.5%), 93.7% (133/142, 95% CI 89.7-97.7%), and 92.6% (138/149, 95% CI 88.4-96.8%). With tetracycline, they were 87.2% (136/156, 95% CI 81.9-92.4%), 95.3% (122/128, 95% CI 91.7-99.0%), and 90.7% (136/150, 95% CI 86.0-95.3%). Amoxicillin-, tetracycline-, and metronidazole-resistant rates were 8.3, 1.0, and 87.8%, respectively. Non-inferiority was confirmed (P<0.025). Metronidazole resistance did not affect the efficacy of either therapy. Compliance was greater and moderate and severe adverse events were less among those receiving amoxicillin than those receiving tetracycline.The novel bismuth-containing quadruple therapy with metronidazole and amoxicillin is an alternative to classical bismuth quadruple therapy for H. pylori rescue treatment as it provides similar eradication with superior safety and compliance.