It is possible to apply machine learning, uncertainty management and paraconsistent logic concepts to the design of a paraconsistent learning system, able to extract useful knowledge even in the presence of inconsistent information in a database. This paper presents a decision tree-based machine learning technique capable of handling inconsistent examples. The intention is to define a model able to handle databases with a large quantity of inconsistent examples. The model obtained is evaluated and compared with the C4.5 algorithm in terms of classification accuracy and size of the trees generated. As will be observed, in most situations where high rates of inconsistent examples were found, this presented better results when compared to the C4.5 algorithm.