Aim -Little is known about the long-term course of patients explored by push enteroscopy for gastrointestinal bleeding of obscure origin. This study aimed to determine the diagnostic yield and the therapeutic impact of enteroscopy, the rate of rebleeding and predictive factors of rebleeding in these patients. Patients and methods -One hundred nineteen patients underwent push enteroscopy for overt bleeding (N = 66) or anemia (N = 53). Results -Enteroscopy was positive in 42%of patients (colon 17%, stomach 13%, small bowel 12%) and diagnosed arteriovenous malformations in two-thirds of patients. Twentyfive additional diagnoses were established during the 2-month follow-up. Treatment was definitive in 13%of patients, without recurrent bleeding. Rebleeding occurred in 45%of patients, and was more frequent when a lesion was visualized (73%vs 28%after 5 years, P = 0.02). In multivoriate analysis, a lesion visualized by enteroscopy was the only independent predictive factor. Conclusion -Enteroscopy is not a high-performance diagnostic tool for obscure gastrointestinal bleeding and enables definitive treatment in less than 15%of patients. Aim -Little is known about the long-term course of patients explored by push enteroscopy for gastrointestinal bleeding of obscure origin. This study aimed to determine the diagnostic yield and the therapeutic impact of enteroscopy, the rate of rebleeding and predictive factors of rebleeding in these Patients. Patients and methods - One hundred nineteen patients underwent push enteroscopy for overt bleeding (N = 66) or anemia (N = 53). Results -Enteroscopy was positive in 42% of patients (colon 17%, stomach 13%, small bowel 12%) and diagnosed arteriovenous malformations in two-thirds of patients. Twentyfive additional diagnoses were established during the 2-month follow-up. Treatment was definitive in 13% of patients, without recurrent bleeding. Rebleeding occurred in 45% of patients, and was more frequent when a lesion was visualized (73% vs 28% after 5 years, P = 0.02). In multivoriate analysis, a lesion visualized by enteroscopy was the only independent predictive factor. Conclusion-Enter oscopy is not a high-performance diagnostic tool for obscure gastrointestinal bleeding and can definitive treatment in less than 15% of patients.