With　the　help　of　eye-tracker,　the　eye　movement　features　on　the　text　message　from　traffic　sign　are　analyzed　based　on　the　word　list　search　task,　in　which　words　with　different　length　were　arranged　in　eight　orientations　and　six　spacing　respectively.　The　fixation　duration　under　different　recognition　mode　is　calculated　by　using　statistical　analysis.　The　saccade　speed　of　different　orientations　and　spacing　is　identified　with　the　help　of　response　surface　method.　The　comparison　is　made　between　the　average　fixation　duration　and　the　average　saccade　time　and　the　results　show　that　the　average　saccade　time　is　just　the　one-fifth　of　average　fixation　duration.