Although some studies have suggested fewer venous complications are associated with midazolam than with diazepam for endoscopic procedures, this variable has not been well documented. We prospectively evaluated the incidence of venous complications after intravenous injection of diazepam or midazolam in 122 consecutive patients undergoing colonoscopy and esophagogastroduodenoscopy. Overall, venous complications were more frequent with diazepam (22 of 62 patients) than with midazolam (4 of 60 patients) (p < 0.001). A palpable venous cord was present in 23% (14 of 62) of patients in the diazepam group, compared with 2% (1 of 60 patients) in the midazolam group (p < 0.002). Pain at the injection site occurred in 35% (22 of 62) of patients in the diazepam group compared with 7% (4 of 60 patients) in the midazolam group (p < 0.001). Swelling and warmth at the injection site were not significantly different between the two groups. Smoking, nonsteroidal anti-inflammatory drug use, intravenous catheter site, dwell time of the needle, alcohol use, and pain during the injection had no effect on the incidence of venous complications.