Recent emphasis on cutting the costs of psychiatric care and the possibility that reimbursement for psychiatric services will one day be based on diagnosis-related groups has stimulated debat about the proper length of psychiatric hospitalization. The authors review the literature on length of stay, focusing primarily on studies of the relationship between various patient and environmental variables and length of stay and studies comparing the outcomes of long and short hospitalizations. They conclude that diagnosis alone is not an accurate predictor of length of stay but may have predictive ability when combined with other data. Most studies found no differences in the outcomes of short and long hospitalizations. The authors identify numerous avenues for further research and are optimistic that a policy governing length of stay is within reach.