In recent years, providing care for individuals with severe mental illness has consumed increasing state and federal financial resources, with State Medicaid systems bearing the heaviest burden. Managed care strategies have been initiated by public mental health systems as a mechanism to control expenses. The state of Maryland implemented a mental health carve-out on July 1, 1997. The purpose of the present study is to describe the effects of the carve-out on access-to-care for individuals using emergency department services in one suburban health care system. Data for this study included all episodes of emergency crisis care in pre-implementation (1996-1997) and post-implementation (1998-1999) time periods. These data were examined within the context of the Behavioral Model of Health Service Use (Andersen, 1995) to describe the interrelationships among external environment, predisposing characteristics, and enabling resources on use of health services. Use of health services was operationalized by four outcomes: disposition, length of stay, number of visits, and recidivism. There were a total of 2986 episodes, initiated by 1928 individuals. Logistic regression demonstrated that when controlling for predisposing characteristics and enabling resources, the likelihood of inpatient admission did not change after initiation of the program. Moreover, there was not a significant change in the number of emergency visits. The assessment of recidivism demonstrated that only psychotic disorders (a predisposing characteristic) were a significant predictor of 30-day repeat visits. Multiple regression models examining the impact of the carve-out on length of stay demonstrated a significant increase in the emergency department length of stay (F = 5.47, p = .05) following the implementation of the carve-out. While benefits associated with improved coordination of services might be expected with the implementation of the carve-out, there was not a change in inpatient admissions, number of emergency visits, or recidivism. Additionally, there was a significant increase in the amount of time required to assess patients and to provide an appropriate disposition. The limited study sample and data prohibit generalizability. Considering that evaluations of mental health carve-outs are limited, this study reflects that anticipated benefits have not been experienced in emergency departments.