This paper presents a pair of early warning models - the logistic regression and the Cox proportional hazards model - and apply them to the prediction of bank failures in Brazil between 1995 and 1998. These models basically produce estimates of the probability that a bank with a given set of characteristics will survive longer than some specified length of time into the future. In addition, these models point out which were the characteristics that most contributed to the bank insolvency. The models identify both failed and healthy banks with a high degree of accuracy. Furthermore, a large proportion of banks that subsequently failed are flagged as potential failures in periods prior to their actual demise. This results demonstrate that reasonably accurate early warning models can be built and maintained at relatively low cost. The use of these models could be of great benefit as an additional instrument to the banking supervision activity.