Similarity　search　of　trajectory　is　the　basis　of　trajectory　clustering　and　pattern　discovering.　The　movement　of　the　moving　objects　is　usually　constrained　by　a　certain　network.　The　popular　methods　for　measuring　the　similarity　of　trajectory　on　general　network　usually　first　transform　the　trajectories　into　a　sequence　of　network＇s　node　or　link　ID,　and　then　apply　the　Longest　Common　Subsequence　(LCSS)　algorithm　to　measure　the　transformed　trajectory.　In　this　paper,　we　aim　to　measure　the　similarity　of　trajectories　on　road　network,　which　has　special　characteristics　compared　with　general　network.　We　proposed　to　transform　the　original　trajectories　into　a　series　of　route　object　using　linear　reference　method　rather　than　node/link　ID　series.　Then　we　modified　the　LCSS　to　enable　it　to　compare　the　route　object　series.　The　advantage　of　our　approach　is　that　it　reduces　the　length　of　transformed　trajectory,　thus　improving　the　efficiency　of　LCSS,　which　is　sensitive　to　the　symbol　length.　We　evaluate　our　method　based　on　a　dataset　generated　by　Network-based　Generator　of　Moving　Objects.　Our　experimental　results　shows　the　proposed　method　has　a　better　performance　than　traditional　methods.　Copyright　©　2012　Binary　Information　Press.