BACKGROUND: The patterns of bleeding following endoscopic sphincterotomy (ES) and their predictive value for subsequent bleeding are poorly understood. Similarly, the efficacy and side effects of epinephrine (E) injection for persistent bleeding have not been well studied. METHODS: Over a 44-month period, all patients undergoing ES were prospectively assessed and followed-up. The character of bleeding (pulsatile, oozing, trickle, none) was recorded immediately, 5 minutes following ES and at the completion of the procedure. Patients with persistent bleeding at the time the procedure was completed (5 minutes or greater) received E injection(s) (1:10,000 concentrations) into the bleeding point with a sclerotherapy needle. ES was performed in all patients with a single electrosurgical generator Valleylab (Force 1B) using pure cutting current. RESULTS: 506 patients (68% females, mean age 54 years) who underwent 550 ES were studied. Bleeding patterns immediately following ES were: 6% pulsatile, 42% oozing, 27% trickle, and 24% none. E (median 0.5 cc; range 0.5-4 cc total) was injected during 79 procedures (14%); none of these patients had complications nor delayed bleeding. For all patients, delayed bleeding occurred in 8 (1.6%, 95% CI 0.57-0.0269); of these 8 delayed bleeders, 1 had no bleeding after ES, and only 1 had any bleeding at 5 minutes. The only variable associated with bleeding after ES was abnormal labs (thrombocytopenia, elevated creatinine concentration, hypoprothrombinemia). CONCLUSIONS: The pattern of bleeding following ES may not predict the risk of late bleeding. Abnormal labs are associated with visible bleeding. Epinephrine injection is safe and appears to provide effective hemostasis.