The clustering of Web usage sessions based on the access patterns is studied. Access patterns of Web users are extracted from Web server log files, and then organized into sessions which represent episodes of interaction between the Web users and the Web server. Using attribute-oriented induction, the sessions are then generalized according to a page hierarchy which organizes pages based on their contents. These generalized sessions are finally clustered using a hierarchical clustering method. Our experiments on a large real data set show that the approach is efficient and practical for Web mining applications.