This study's purpose was to develop a tool that detects, quantifies, and assigns causes for medically unnecessary hospital delays and use it to describe the epidemiology of delays at a teaching tertiary care hospital. Based on observational data, a taxonomy of delays was constructed that included nine major categories and 166 subcategories. This formed the basis for an instrument for detecting inefficiency in hospital care: the Delay Tool. Initially designed for real-time concurrent assessment, in retrospective use it was also reliable, requiring about 6 minutes per medical record. In using the Delay Tool over a 6-month period on general internal medical and gastrointestinal services, it was discovered that 30% of 960 patients experienced delays, the average length of which was 2.9 days. This represented 17% of all hospital days. The most frequent causes of delays were scheduling of tests (31%), unavailability of post-discharge facilities (18%), physician decision-making (13%), discharge planning (12%), and scheduling of surgery (12%). Because of the longer lengths of the delays involved with awaiting postdischarge facilities (primarily nursing home beds), this was the most important cause of delays and represented 41% of all delay days. The general medicine and gastrointestinal services had significantly different distributions of delay types related to their different kinds of patients and care. The Delay Tool should be helpful in addressing hospital, and hospital-related, inefficiencies in health care delivery.