Traffic　congestion,　which　is　represented　by　stochastic　short-term　traffic　interference　of　varying　duration　and　frequency,　is　a　key　influencing　factor　of　traffic　characteristics　and　quality　of　service　of　urban　street　network.　The　two-fluid　(the　vehicles　in　traffic　flow　are　divided　into　moving　vehicles　and　stopped　vehicles)　model　is　able　to　describe　the　average　operating　performance　of　urban　traffic,　but　it　is　difficult　to　collect　individual　vehicles　data　in　the　network　for　practical　application　of　the　model　in　urban　area.　The　sampling　strategies　on　data　collection　based　on　probe　vehicles　are　stated　in　the　paper　and　trip　histories　of　probe　vehicles　are　aggregated　across　link-based　microtrip　in　order　to　estimate　the　model　parameters.　The　algorithm　of　cost　function　is　applied　to　map-matching　with　the　help　of　GIS　(Geographic　Information　System).　The　critical　speed　for　identifying　the　vehicle　state　is　determined　through　sensitivity　analysis.　The　field　experiments　showed　that　the　application　of　probe　vehicle　data　is　reliable　to　assess　the　network　performance　and　traffic　condition　of　urban　street　network.　©　2007　IEEE.