OBJECTIVE: Emergency surgery after unsuccessful endoscopic therapy for bleeding duodenal ulcer has been reported to have a high mortality. Transcatheter arterial embolization (TAE) of the gastroduodenal artery is an alternative strategy when endoscopic therapy fails. This study is a retrospective analysis comparing these two treatment strategies. MATERIAL AND METHODS: Patients who underwent TAE (n = 24) or open surgery (n = 50) after unsuccessful endoscopic therapy for bleeding duodenal ulcers at two university hospitals between 2000 and 2007 were compared. Mortality, morbidity, length of hospital stay, age, number of endoscopic interventions and acute physiology and chronic health evaluation (APACHE) II score were evaluated. RESULTS: The groups were comparable concerning gender and length of hospital stay. The mean age (69.6 +/- 16.1 versus 61.9 +/- 14.1 years; P = 0.043), APACHE II score (17.0 +/- 5.1 versus 12.8 +/- 5.7; P = 0.004) and number of gastroscopies (P = 0.009) were significantly higher in the embolization group. Five patients (20.8%) died in the embolization group compared to 11 (22%) in the surgery group. However, mortality in high-risk patients (APACHE II score >or= 16.5) was lower in the TAE group (23.1% versus 50.0%; P = 0.236). Method-related as well as other complications were not significantly different between the two groups. There was, however, a higher re-bleeding rate in the TAE group. CONCLUSIONS: TAE of the gastroduodenal artery appears to be a safe alternative when endoscopic therapy for bleeding duodenal ulcer fails, at least in high-risk patients. The role of TAE in low-risk patients with bleeding from duodenal ulcer needs to be defined by means of a prospective controlled trial.