In　this　paper,　the　packet　traffic　flow　on　weighted　scale-free　networks　is　guided　based　on　local　link　information　and　a　single　adaptive　preferential　factor　alpha　which　characterizes　the　preferential　probability　in　choosing　links　to　forward　packets:　P-l　-＞　i　=　w(li)(alpha　li)/Sigma(j)w(lj)(alpha　lj).　The　adaptive　preferential　factor　alpha　increases　when　the　queue　length　of　the　link　is　smaller　than　the　capacity,　and　decreases　otherwise.　The　capacity　of　links　is　set　to　be　the　weight　of　the　link.　It　is　shown　by　simulations　that　using　adaptive　preferential　factor　alpha　could　both　reduce　the　packets　delivering　time　and　increase　the　capacity　of　huge　communication　networks　compared　with　using　fixed　value　of　alpha.