In the last few years the size and coverage of Wikipe- dia, a freely available on-line encyclopedia has reached the point where it can be utilized similar to an ontology or tax- onomy to identify the topics discussed in a document. In this paper we will show that even a simple algorithm that exploits only the titles and categories of Wikipedia articles can characterize documents by Wikipedia categories sur- prisingly well. We test the reliability of our method by pre- dicting categories ofWikipedia articles themselves based on their bodies, and by performing classification and cluster- ing on 20 Newsgroups and RCV1, representing documents by their Wikipedia categories instead of their texts.