Advanced age and comorbidity as well as gastrointestinal (GI) disease contribute to the increased mortality after upper GI endoscopy in inpatients when compared to outpatients. The aim of this study was to measure comorbidity in inpatients undergoing endoscopy using the Acute Physiology and Chronic Health Evaluation (APACHE) II severity of disease classification and to assess the usefulness of the APACHE II system in predicting outcome. During a 10-week period, 155 consecutive inpatients undergoing upper GI endoscopy were prospectively scored using APACHE II. They were followed up for 30 days, the measured endpoint being death. Of these, 92 (59%) inpatients were admitted with GI hemorrhage, 14 (9%) were admitted for other reasons but subsequently bled, and 49 (32%) were endoscoped for reasons other than bleeding. The mean (SEM) APACHE II score in patients with GI bleeding was 8.0 (0.5), and in patients without bleeding was 6.5 (0.6;p= 0.07). Eighteen patients (12%) died within 30 days of endoscopy. APACHE scores were higher at 10.5 (1.2) in patients who died, compared to 7.1 (0.4) in those who lived (p< 0.01). Increased acute physiology scores led to this difference. Age and chronic health scores were similar in both groups. In the 18 patients who died, 9 had GI bleeding and their mean APACHE score was 13.8 (1.5); 9 had been endoscoped for other reasons and had a lower score of 7.2 (1.3;p< 0.01). These latter 9 deaths amounted to a 18% mortality in the nonbleeding group, which was greater than expected. APACHE II scores can help predict poor outcome in inpatients referred for endoscopy. However, the APACHE II system has limitations and failed to identify (by means of a high score) some patients without GI bleeding who subsequently died. A tool to measure comorbidity, such as the APACHE II system, is necessary when comparing groups of patients in different settings.