This study compares the effects of two different benzodiazepines used for conscious sedation during combined upper gastrointestinal endoscopy (EGD) and colonoscopy. Subjects were assessed for their degree of analgesia and amnesia for the procedure, prior experience with endoscopy, and willingness to undergo another similar procedure should such be necessary. The patients were randomized single blind to receive either midazolam or diazepam for their preprocedure sedation. The amount of preprocedure sedation utilized was determined by titration of the dose to achieve slurring of speech. Prior to receiving either agent, the subjects were shown a standard card containing pictures of 10 common objects, were asked to name and remember them, and were told they would be "quizzed" (at 30 min and 24 hr) after being sedated for their recollection as to the objects pictured on the card. Each subject filled out a questionnaire addressing their perceived discomfort during the endoscopic procedure and their memory of the procedure 24 hr after the procedure. Sixty-three percent of the midazolam-sedated subjects reported total amnesia for their colonoscopy vs 20% of diazepam-sedated patients (P less than 0.001). Fifty-three percent of midazolam-sedated patients reported total amnesia of their upper gastrointestinal endoscopy vs only 23% of diazepam-sedated subjects (P less than 0.05). The midazolam-sedated subjects reported experiencing less pain with both upper gastrointestinal endoscopy (P less than 0.05) and colonoscopy (P less than 0.001) than did the diazepam-sedated group. Most importantly, the midazolam group was more willing to undergo another similar endoscopic procedure should they be asked to do so by their physician (P less than 0.05).(ABSTRACT TRUNCATED AT 250 WORDS)