Over　the　past　15　years,　Global　Positioning　System　(GPS)　technology　has　been　frequently　used　as　a　tool　to　detect　potential　earth　mass　movements　and　to　track　creeping　landslides.　In　this　study,　we　investigated　4　years　of　continuous　GPS　data　(September　2006-July　2010)　recorded　at　a　landslide　site　in　Alaska.　This　GPS　station　(AC55)　was　installed　on　an　un-identified　creeping　site　by　the　Plate　Boundary　Observatory　(PBO)　project,　which　was　funded　by　the　US　National　Science　Foundation.　The　landslide　moves　with　a　steady　horizontal　velocity　of　5.5　cm/year　toward　NEE　15A　degrees　and　experiences　a　steady　subsidence　of　2.6　cm/year.　There　is　a　considerable　correlation　between　annual　snow　loading　and　melting　cycles　and　seasonal　variations　in　the　landslide　displacements.　The　seasonal　movements　vary　year　to　year　with　an　average　peak-to-trough　amplitude　of　1.5　and　1.0　cm　in　vertical　and　horizontal　directions,　respectively.　This　study　addresses　three　challenging　issues　in　applying　GPS　for　landslide　monitoring　in　tectonically　active　and　cold　regions.　The　three　challenges　include　(1)　detecting　GPS-derived　positions　that　could　be　contaminated　by　the　snow　and　ice　accumulated　on　GPS　antennas　during　cold　seasons,　(2)　establishing　a　stable　local　reference　frame　and　assessing　its　accuracy,　and　(3)　excluding　local　seasonal　ground　motions　from　GPS-derived　landslide　displacement　time　series.　The　methods　introduced　in　this　study　will　be　useful　for　GPS　landslide　monitoring　in　other　tectonically　active　and/or　cold　regions.