OBJECTIVES: Providers who have had favorable cost and quality trends may be more likely to form Accountable Care Organizations (ACOs) because they expect to profit from changes that are already underway within their organization. We examine trends in hospital cost per discharge and in-hospital mortality rates among hospitals that formed ACOs and those that did not, incorporating several years of data preceding ACO formation. METHODS: We compared growth rates in cost per discharge and in-hospital mortality rates for select conditions. Data were from 2008 to 2011 (pre-ACO) and 2011 to 2012 (post-ACO) for hospitals that did and did not implement ACOs. We also explored whether there were distinct trends based on ACO leadership structure. RESULTS: Between 2008 and 2011, the average rate of growth in cost per discharge for hospitals that adopted ACOs was less than one-third of the rate among hospitals that remained unaffiliated (0.59% vs 2.02%). Among ACOs in which the hospital assumed a leadership role, mean cost per discharge declined during the pre-ACO period at an average rate of 0.55% for hospital-led ACOs and 1.52% for jointly-led ACOs. Cost per discharge during the post-ACO period grew at a rate of 1.99% among ACO hospitals and 1.02% among non-ACO hospitals. Hospital-led ACOs experienced a 1.95% increase in cost per discharge between 2011 and 2012, while cost per discharge among jointly-led ACOs fell by only 1.27%. Analysis of in-hospital mortality rates did not reveal persistent trend differences. CONCLUSIONS: Hospitals that adopted the ACO model had more favorable cost trends between 2008 and 2011 than hospitals that did not adopt the model, which suggests non-random selection of providers opting to participate in ACO initiatives. In the post-ACO adoption period, hospitals that were part of jointly-led ACOs had the lowest cost growth, suggesting that this ACO structure may be the most effective.