In many real world problems it is often difficult to find dependencies between the variables of a process or more general of a system, dependencies which can be used for controlling a plant, forecasting a value or classifying a group of objects into pre-defined classes. Since in many cases, analytic dependencies are unknown or very difficult to set up, the formulation of dependencies with the help of fuzzy rules offers a useful alternative. This paper presents the combined use of a fuzzy rule generation method and a data mining technique for financial risk assessment. The case of business failure is considered here and the classification of the firms into two classes is sought. Initially, a method for the generation of fuzzy rules is used. Then these rules are imported to a data mining technique so as the firms can be classified into as bankrupt or non-bankrupt. The fuzzy method supports the discovery of relevant dependencies by the automatic generation of if/then rules on the basis of expert knowledge, while the data mining technique, with the help of a fuzzy rule-based classifier, assigns an object to different classes on the basis of various different characteristics (financial ratios). Finally, a thorough comparison with discriminant analysis, logit and probit analysis is performed based on the same sample.