The purpose of this study was to determine the rates and causes of revisits to the emergency department (ED) of a 3,500-bed referral hospital, to identify areas for improvement, and to generate baseline data for the development of a computerized, automatic monthly audit system. We identified all patients returning within 72 hours of their initial ED visits, from 1 July 1995 to 30 June 1997, and monthly revisit rates were calculated. To determine the reasons for revisits, two independent reviewers examined the charts of revisit cases from 1 July 1996 to 30 June 1997. A one-in-three sampling method was used to select charts. A total of 485 revisit charts were reviewed. The monthly revisit rates ranged from 1.32% to 2.38%, with no particular seasonal or event-specific pattern. Most revisits were attributed to disease factors (79.0%). Those felt to be medical errors only accounted for 7.8% of the revisits but led to a higher hospital admission rate (73.7%) subsequently. By contrast, the overall hospital admission rates for revisit patients (36.5%) and all ED patients (36.2%) were similar. We suggest setting baseline monthly ED revisit rates at 2% for future computer-programmed audit filters. While this study indicates that most revisits are disease-related, further prospective studies are needed to evaluate the most common and serious causes of revisits to see if improvements can be made.