Road　traffic　is　a　random　and　dynamic　system　while　it　is　difficult　to　describe　the　traffic　flow　characteristics　correctly,　through　the　existing　non-probability　parameters.　Evaluating　the　road　network　with　such　parameters　will　produce　difficulty　in　traffic　system　management　and　planning.　This　paper　guided　by　the　theories　of　probability　and　traffic　flow　with　primary　consideration　of　random　arrival　of　vehicle,　aiming　at　the　shortcomings　of　the　past　evaluation　method,　the　unblocked　thresholds　in　the　urban　roads　of　different　levels　are　determined　according　to　field　survey　of　large　numbers　of　drivers.　On　this　basis,　the　parameter　values　of　the　BPR　and　the　regression　function　are　calibrated　combined　with　road　conditions　in　Tianjin　of　China　and　this　paper　gets　the　probability　model　of　unblocked　flow　based　on　the　random　arrival　of　vehicle.　Besides,　the　model　is　verified　with　simulation　software　Paramics.　It　is　anew　idea　about　the　calculation　method　of　unblocked　reliability.　©　2013　Asian　Network　for　Scientific　Information.