The　total　mileage　of　highways　in　China　has　ranked　the　second　in　the　world　and　traffic　accidents　have　become　a　potential　safety　problem,　which　threaten　people＇s　lives.　Panel　data　model　as　a　new　tool　is　introduced　to　analyze　crashes.　Panel　data,　which　combines　cross-sectional　and　time-series　characteristics,　allows　researchers　to　construct　and　test　realistic　behavioral　models　that　cannot　be　identified　using　only　cross-sectional　or　time-series　data.　This　paper　introduces　the　modeling　process　of　fixed　effects　model　and　random　effects　model.　These　models　are　applied　to　the　Jingjintang　Highway.　Variables　are　divided　into　four　categories　including　roadway　variables,　traffic　variables,　indicator　for　different　areas,　and　accidents　variable.　The　two　models＇　results　are　compared　and　the　fixed-effects　model　was　chosen　to　describe　the　relationship　between　accidents　and　other　factors　after　the　Hausman　test.　The　fixed　effects　model　was　chosen　since　it　could　identify　unobserved　factors　and　segments　with　high　safety　risks.　Panel　data　model　as　a　new　statistical　tool　will　be　widely　used　in　the　transportation　analysis　field.　©　2009　ASCE.