The　value　of　run-time　application　power　consumption　in　the　full　system　is　a　prerequisite　for　energy　optimization.　Though　past　studies　have　demonstrated　the　use　of　performance　events　for　building　single　component　(CPU,　Memory,　Disk)　power　model,　we　propose　a　method　to　estimate　application　power　consumption　in　the　full　system　with　performance　events.　In　this　paper,　we　identify　different　stress　on　components　from　CPU-intensive,　memory-intensive　and　I/O-intensive　applications.　Combining　these　and　the　＂trickle-down＂　effect　of　performance　events,　the　feasibility　of　estimating　application　power　consumption　with　a　few　performance　events　is　demonstrated.　Through　measurement　of　actual　systems　running　test　sets,　these　models　are　shown　to　have　an　average　error　of　less　than　six　percent　across　the　considered　workloads.　Besides,　we　demonstrated　that　these　models　can　be　applied　to　real-world　applications.　Moreover,　power　models　proposed　in　this　paper　can　provide　software　developers　or　system　designers　with　visible　power　behaviors　for　the　applications.　Thus,　software　developers　can　make　decisions　or　change　code　to　develop　more　energy-efficient　software.