OBJECTIVE: To test the hypothesis that anxiety, depressive, or cognitive disorders are associated with an increase in length of stay of physical rehabilitation inpatients. DESIGN: Secondary analysis of a 1-yr prospective data recording. Three treatment and rehabilitation centers in the Canton of Vaud (Switzerland). Ninety-five percent of inpatients admitted from November 15, 1990, to November 14, 1991, agreed to participate. Apart from length of stay, data consisted of demographic and medical data results from the Hospital Anxiety and Depression Scale, Mini-Mental State Score, and Functional Autonomy Measurement System. Multivariate linear regression was used in the analysis. RESULTS: The presence of anxiety or depression altered length of stay in a bivariate analysis, although all effects disappeared in a multivariate approach. Factors that had an independent association with length of stay were gender, length of stay in an acute care hospital before hospitalization, treatment and rehabilitative centers, Functional Autonomy Measurement System mobility score, and Functional Autonomy Measurement System Activities of Daily Living score. Results concerning the association between cognition abilities and length were similar. CONCLUSIONS: Our results recognize that an influence of psychiatric disorders acted on length of stay through a relationship between the psychiatric status and the control variables. If mental state influences physical state, then early intervention studies are desirable. If somatic state induces mental alterations, then interventions directed toward the psychiatric sphere will bring mostly qualitative benefits (amelioration of well-being without remarkable effects on length of stay). [Authors]