OBJECTIVES: To describe and compare both overuse and underuse of diagnostic upper gastrointestinal endoscopy in different settings. DESIGN: Merging of data from three prospective observational studies. The appropriateness and necessity of indications for gastroscopy were evaluated using explicit criteria developed by a standardized expert panel method (RAND-UCLA). Inappropriate endoscopies represent overuse. Necessary indications not referred for the procedure constitute underuse. SETTING: Three primary care outpatient clinics, 20 general practices, three gastroenterology practices, two district and one university hospitals. SUBJECTS: A third of the collective were consecutive ambulatory patients with upper abdominal complaints, whereas the other two thirds were ambulatory and hospitalized patients referred for the procedure. MAIN OUTCOME MEASURES: Proportions of overuse and underuse in the different settings. RESULTS: A total of 2885 patients were included (mean age, 49 years, 52% male, 2442 outpatients), 1858 patients underwent > or = 1 endoscopy. Among 2086 endoscopies, 805 (39%) were inappropriate, most of which were performed for dyspepsia (83%). Overuse was higher in young, foreign, female patients and lower in inpatient settings, the latter reflecting a different distribution of presenting symptoms. Among 1646 patient visits in primary care, overuse represented 148 endoscopies (9%). Underuse was identified in 104 of the same patient visits (6%) and was higher as patient age increased; there were no significant differences between men and women. CONCLUSIONS: Rates of overuse and underuse depend mainly on case presentation and patient characteristics. Both over- and underuse should be addressed to maintain and improve quality of care.