When　accessing　websites,　users　usually　leave　a　lot　of　access　information,　which　can　be　mined　reasonably　to　help　the　managers　of　website　to　get　accessing　patterns　of　users.　This　article　first　introduces　the　preprocessing　procedure　of　web　logs,　which　includes　the　tasks　of　data　cleaning,　Data　Discretization　and　their　implementation.　On　the　basis　of　preprocessing　the　analysis　method　of　requested　resource　＇URL＇　and　＇referrer＇　which　is　the　webpage　before　users　browse　the　URL　in　web　log　by　the　use　of　association　rules　is　proposed　to　find　the　accessing　patterns　of　users.　Finally　the　experiment　is　accomplished.　The　result　shows　that　the　method　is　feasible,　and　it　can　help　the　manager　in　making　decisions　about　the　analysis　of　website　users＇　behavior　and　the　optimization　of　website.　©　2011　IEEE.