In the paper we investigated the problem of banks’ financial state forecasting, carried out a comparative analysis of different methods for bank bankruptcy prediction, and determined the financial indicators that can be used in the models. Both fuzzy logic and crisp approaches has been used to solve the forecasting problem. We have developed a software which allows to apply the fuzzy neural networks ANFIS and TSK, fuzzy group method of data handling as well as crisp methods: logit, probit model, regression models, Kromonov method and method of multilevel aggregate index of bank state to predict the financial state of banks. The software also allows conducting a comparative analysis of the received results. We used the financial statements of Ukrainian banks and largest European banks as input parameters.