In this paper, length of stay for psychiatric inpatients (LOS) is discussed in terms of controlled studies of hospitalization, research on patient factors contributing to LOS, and variation in mean or median LOS between institutions and within units. Although the multiplicity of variables involved precludes ready generalization, it can be said that open-ended hospital stays, above 28 days, have not been demonstrated to improve outcomes. However, the complexity of specific circumstances makes it unrealistic to impose rigid uniform standards within or across units. Medically managed units with a director as the locus of decision making may be in a better position to control LOS. Research into relationships between different treatment milieus and outcomes for specific diagnoses is suggested as a promising avenue for further research into cost effectiveness of hospitalization.