Despite evidence from controlled trials that routine diagnostic endoscopy does not improve the outcome of patients with acute upper gastrointestinal bleeding (UGIB), available information suggests that it is widely used in this situation. Proponents of endoscopy justify its use by citing several factors, such as knowledge of the bleeding source, that were not addressed in the clinical trials. It seems, therefore, that either diagnostic upper endoscopy is being overutilized or that the data provided by the available clinical trials are inadequate to definitively assess its proper role in these patients. The purpose of this study is to assess the role of diagnostic endoscopy in adult patients with uncomplicated UGIB using the analytic hierarchy process (AHP), a multicriteria decision making technique. We hypothesize that upper endoscopy is being overutilized. To test this hypothesis we will interview primary care physicians and patients who have recently had an uncomplicated acute UGIB. The interviews will elicit information needed to complete individual AHP analyses for each subject to determine the patient care strategy most consistent with their interpretations of the relevant data and judgements regarding the rela- tive importance of 5 patient management goals, including knowledge of the bleeding site. We will then compare the proportion of individuals in each group who, through AHP analysis, identify routine endoscopy as the preferred management strategy with the current rate of endoscopy at our hospital (85%). A significant difference between these 2 rates would confirm our hypothesis and indicate that primary care physicians and patients at our hospital do not value information about the bleeding site highly enough to justify routine endoscopy for most of these patients. This finding may apply to other diagnostic testing decisions. Regardless of the outcome, this study will demonstrate a new method for assessing diagnostic technology and provide information about patients' and physi- cians' attitudes regarding the use of diagnostic testing in acute UGIB.