Abstract In this paper we describe MultiModal WebRemUsine, a tool for remote usability evaluation of Web sites that considers data regarding the user behaviour coming from multiple sources. The tool performs an automatic evaluation of the usability of the considered Web site by comparing such data with that contained in the task model associated with the pages (which describes the expected behavior of the user). The results of the analysis are provided along with information regarding the user behavior during the task performance. Using such information, evaluators can identify problematic parts of the Web site and make improvements, when necessary. An example of application of the proposed method is also discussed in the paper.