Background: The use of push enteroscopy to evaluate patients with obscure gastrointestinal bleeding has increased in recent years, and diagnostic yield has been reported to be 50% to 65%. This yield may be an overestimate of accuracy, as some lesions found during enteroscopy are within reach of a standard endoscope. Methods: Ninety-five patients underwent push enteroscopy for obscure gastrointestinal bleeding. There were 58 men and 37 women with a mean age of 67 years (range 32 to 93 years). Diagnostic yield and patient outcome were assessed. Results: A suspected source of bleeding was found in 39 of 95 patients (16 of these patients had endoscopic treatment of their lesions). Proximal lesions (at or above the main duodenal papilla) accounted for 25 of 39 sources (64%), including Cameron ulcers and arteriovenous malformations of the stomach/proximal duodenum. Distal lesions accounted for 14 of 39 sources (36%) with arteriovenous malformations (n = 10) being most common. Patients who underwent some form of treatment (medical, surgical, or endoscopic) because of an enteroscopic finding had a statistically better outcome than patients without a lesion (73% vs. 47%, p < 0.05). Conclusions: Push enteroscopy identified a presumed bleeding source in 41% of patients with obscure gastrointestinal bleeding. However, 64% were within reach of a standard endoscope. Repeat standard endoscopy should be considered before push enteroscopy for obscure gastrointestinal bleeding, and during enteroscopy meticulous attention should be given to the proximal gastrointestinal tract in addition to the distal duodenum and jejunum. (Gastrointest Endosc 1998;47:88336-6.)