A quasi-experimental method was developed to evaluate the cost-effectiveness of a public system of 24-hour acute psychiatric care in Santa Clara County, California, before and after a new treatment setting was introduced. The original system relied on a 54-bed psychiatric unit in a county general hospital; the new system consisted of a 20-bed unit in the general hospital plus a 45-bed nonhospital psychiatric health facility. The study demonstrated that the per diem cost of the psychiatric health facility was approximately 60 percent that of the original general hospital unit, but the average difference in cost per episode between the two systems was only about $25, primarily due to longer lengths of stay in the new system. In addition, patients treated in the new, combined system appeared sicker at discharge than those treated in the old system. The findings suggest the importance of simultaneously evaluating both cost and treatment effectiveness to make sure that one element does not dominate program direction at the expense of the other.