||Decision trees are a well known and widely used method for classification problems. For handling numerical attributes or even for numeric prediction, traditional decision trees based on crisp predicates are not suitable. By using fuzzy sets for the test predicates and for the goal predicates we show how decision trees can be used to handel continuous attributes, too. We further show, how the fuzzy logical inference methods like Mamadani and Sugeno inference can be used to predict numerical values, too. Finally, we will present some applications of fuzzy decision trees and compare the results with other rule base approaches.