The main aim of this paper is to provide general information about emerging risk in Indonesian banks using econometric models. However, it also demonstrates how the models employed can be used to produce “probability scores” for banks thought likely to suffer problems, which can be used as an early warning system in banking supervision. Previous studies in this area have normally been concerned with the assessment of the probability of bank failure, although the assessment of bank condition prior to failure is more important for banking supervisors. Moreover, the number of bank failures recorded in these earlier studies was often very low, causing statistical problems for researchers. In these respects, this paper marks an improvement on previous empirical studies. The paper employs pool data of problem and non-problem banks as dependent variables and financial ratios, which represent various risks in banks, as independent variables. Using an out of sample test, the results show that an appropriately-specified logit model can estimate 87.82% of observations correctly.