BACKGROUND: Reducing length of hospital stay (LOS) is a policy aim for many health care systems and is thought to indicate efficiency. METHODS: A MedLine search was undertaken for articles relating to 'LOS', 'early discharge' or 'patient discharge' between 1983 and 1997 and a selective search was undertaken for material published before 1983. RESULTS: Routine data showed that there were variations in LOS between countries, regions and hospitals. The trends in LOS showed a decrease over time in all regions. Research consistently fails to show an adverse effect on health outcomes of reducing LOS, but there may nevertheless be an ethical or moral minimum LOS. Two recent examples illustrate this. There has been an outcry at some ultrashort stays, for example 'drive-through mastectomy' and 'lunchtime abortion' and these are discussed in the review. CONCLUSIONS: There are a number of reasons for the perceived lack of relationship between LOS and health outcomes. Clearly reducing days of care at the low-intensity end of a hospital stay may not necessarily affect health outcomes. There is a case to be made for tailoring care more exactly to an individual's needs by looking at the actual components of care rather than the place of care--within or outside hospital walls.