In this paper we describe a clustering method that allows the use of graph-based representations of data instead of traditional vector-based representations. Using this new method we conduct content-based clustering of two web document collections. Clustering of web documents is performed to organize the documents with little or no human intervention. Benefits of clustering include easier browsing and improved retrieval speed. In order to measure the performance of our graph-matching approach, we compare it to the popular vector-based k-means method. We perform experiments using different graph distance measures as well as various document representations that utilize graphs. The results with the k-means clustering algorithm show that the graph-based approach can outperform traditional vector-based methods.