Although coercive measures have always been part of the psychiatric armamentarium, the ethical dilemma between the use of a "therapeutic" coercion and the loss of patients' dignity is one of the major controversial issues in mental health research and practice. The aims of the present review are to explore the existing literature on predictors of use of coercive measures and to explore the relationship between coercive measures and patient outcome. A literature search was conducted using MEDLINE, PsychyINFO, Scopus, Web of Knowledge and the Cochrane Database. In all selected papers, references were cross-checked to identify other possible eligible papers. The use of coercive measures was predicted by patients' clinical and socio-demographic features, staff characteristics and ward-related factors. Coercive measures have only a limited impact on patients' clinical and social outcome. At the current level of knowledge, coercion is still a controversial issue in mental health practice. Only few studies with a solid methodology have been carried out. Large multicenter and rigorous studies, with long-term follow-ups, are highly needed.