Journal of International Financial Markets, Institutions and Money, v. 28, pp. 54–65. Abstract: This paper analyses efficiency in Brazilian banks from 1998 to 2010 with a Bayesian dynamic frontier model. This model provides a more structural explanation for the variation in bank inefficiency than that has been presented by previous models, and also allows for cost inefficiency effects. On average, the dynamic frontier results, estimated via the Markov Chain Monte-Carlo simulation, indicate that Brazilian banks improved in terms of efficiency over time. Factors found to be important determinants of cost efficiency include public banks and foreign banks that are statistical insignificants, merger and acquisitions, big banks, deregulation and stressed banks that are statistical significant. Big bank and deregulation are the only variables that decrease costs in the Brazil market. Several Policy implications are derived.