Background The urea breath test (UBT) is one of the most accurate methods of assessing Helicobacter pylori status. The predictive value of the test is, however, uncertain. This study was a serial, prospective analysis of the change over time of UBT values after first-, second- and third-line treatments of patients with failed eradication therapy. Methods. One hundred thirty-four duodenal ulcer patients with persisting H. pylori infection after first-line triple therapy were enrolled in a cross-over manner to receive either pantoprazole (40 mg twice daily), amoxicillin (1000 mg twice daily), and clarithromycin (500 mg) or ranitidine bismuth citrate (400 mg twice daily), metronidazole (250 mg twice daily), and clarithromycin (500 mg twice daily) for 7 days. Forty-one patients with failed second-line treatment were randomized to receive third-line quadruple therapies with pantoprazole + amoxicillin and tetracycline (500 mg four times daily) and either nitrofurantoin (100 mg three times daily) or bismuth subsalicylate (120 mg four times daily). Breath tests were performed 6 weeks after therapy. The delta(CO2)-C-13 values (%omicron) after primary, secondary, and tertiary treatment were analyzed, and the correlation between pretreatment values and the rate of H. pylori eradication was assessed. Results. In patients with successful second-line treatment, UBT values decreased from 12.4%omicron [confidence interval (CI), 9.7-15.7)] to 2.8%omicron (CI, 0.9-2.5) (P = 0.001), and in those with persistent infection, they increased from 13.2%omicron (CI, 7.3-19.1) to 19.2%omicron (CI, 13.4-25.0) (P = 0.03). After a failed quadruple regimen, UBT values increased from 19.3%omicron (CI, 16.2-22.4) to 25.8%omicron (CI, 19.8-312.8) (P = 0.03). The correlation between the pretreatment UBT values and the rate of eradication was negative for both second- and third-line therapies. Conclusions. Serial assessment showed that UBT values after successive treatments showed a marked tendency