AIM Hospital length of stay (LOS) is one of the stratified measures of health-care efficiency and is commonly used to assess performance of psychiatric care. The aim of this study was to identify risk factors for prolonged LOS of psychiatric patients. METHODS We retrospectively analyzed adult psychiatric patients (ICD-10; F00-F99) between April 2012 and March 2014 in the Japanese Diagnosis Procedure Combination database. We examined factors affecting prolonged LOS using multivariable logistic regression analysis. Subgroup analyses of the logistic regression were undertaken according to two diagnostic groups (F20-F29 and F30-F39). RESULTS A total of 34 326 patients admitted to and discharged from psychiatric beds were included. Older age, lower Global Assessment of Functioning score, involuntary commitment, several psychiatric services, certain other patient factors, academic hospitals, public hospitals, and higher density of psychiatric beds were significantly associated with prolonged LOS. Hospital patient volume was significantly associated with shorter LOS. In the subgroup analyses, most of these factors were consistent although some were not associated with prolonged LOS. CONCLUSION Not only clinical factors but also institutional characteristics were associated with prolonged LOS. Our study provided useful information for improvement in psychiatric services and indicated the need to consider the division of roles between healthcare/welfare institutions and psychiatric-related resource allocation. Interventions should be considered for achieving shorter LOS for psychiatric patients.