If well-trained physicians are delivering excellent care using an organized multidisciplinary model, why is a database necessary? This question is best answered by examining three fundamental reasons why a clinical management database has become an essential component of an ICU program. First, a comprehensive approach to improving quality and efficiency in critical care requires examining patient characteristics, ICU practice patterns, and outcomes at a population level. Because of the difficulties of recalling infrequent events and subjective impressions of our own practice patterns, the observations made in the care of individual patients cannot be assembled into an accurate picture of ICU performance. By use of validated severity of illness scoring systems and standardized definitions for events and outcomes, a database provides an objective tool to understand the care delivered. The ability to analyze clinical outcomes and resource utilization simultaneously, however, is the greatest strength of a clinical database. Cost and efficiency must be improved but not at the expense of quality. The second reason for a clinical database is documentation. As physicians are increasingly held accountable for both quality and cost, if improvements in care are not documented, they either did not occur or were done by someone else. A database provides systematic documentation of ICU performance, recording the results of quality-improvement initiatives. The third reason for an ICU database is the opportunity it provides for participation in clinical research. Although cause and effect relationships are difficult to establish from a retrospective database review, the database provides a format for standardized data collection. Once structured data collection has been implemented, the foundation then exists for hypothesis-testing prospective research.