This paper summarizes the results of a search of electronic databases for papers on special observation (SO). Published studies to date about SO are entirely descriptive. No evaluative research appears to have taken place, leaving the procedure based on clinical pragmatism and tradition. Something between 3%-20% of admissions receive some form of SO during their stay and the rate of usage varies widely between wards. SO is used as a method of controlling and containing the most disturbed patients who are considered to be imminently at risk of harming themselves or others. Such patients tend to be younger and suffering from acute psychosis or depression. Which professional staff have the authority to initiate and terminate SO varies from place to place, as does its duration. The financial costs have been crudely assessed and are reported to be very high, perhaps up to 20% of the nursing budget for a hospital. Further variation exists on who is allowed to carry out SO. Nurses frequently make unofficial modifications to the procedure based upon their own individual judgments and assessments, and policies vary widely among hospitals. There is little agreement between authorities on what nurses should do during SO, although there is some evidence that it can, under certain circumstances, be therapeutic. However there is also evidence that nurses find SO stressful and patients dislike it.