The study of machine learning has helped create and refine many types of predictive models. These models have been applied to countless problems, but the importance of banking stability has led many researchers to create models predicting if a bank will be active in future years. Neural networks, support vector machines, and traditional regression models, have been used to predict bank failure, but random forests have not been fully explored. This paper shows random forests can offer prediction power greater than neural networks and logit models when used to predict bank failure. This aligns with work comparing random forests and neural networks in other disciplines. The readability of a random forest offers a distinct advantage over neural networks when used for research purposes.