Word sense disambiguation (WSD) is a technique to find the exact sense of an ambiguous word in a particular context.For example, an ambiguous word ’bank ’,that has two senses ’institution’ and ’river bank’ in different contexts.Proposed system will predict the correct meaning of the ambiguous word in a particular context. The proposed work presents a supervised decision tree based learning approach to word sense disambiguation where a decision tree assigns a sense to an ambiguous word based on the set of positional and contextual features . The current approach proposes a classifier based on the REPTree supervised learning algorithm. As an initial step a set of positional and contextual features are inferred after preprocessing the collected database. Based on this features, tree is constructed. Finally, classifier will more precisely classify the new test entries. Weka machine learning tool is used for word sense disambiguation.