To　construct　an　artificial　system　with　multi-Agents,　it　is　obvious　that　some　important　factors,　such　as　different　agent＇s　interests,　limited　resources　and　so　on,　will　inevitably　lead　to　conflict.　To　reduce　conflict,　it　is　found　that　effective　competition　with　negotiation　among　multi-Agents　can　improve　overall　performance.　Thence,　this　paper　proposes　a　new　Multi-Agents　Competition　Model　with　Negotiation,　which　improves　the　forecast　accuracy　of　opponent＇s　competing　strategies　with　negotiation　information　and　shortens　negotiation　time　effectively　depending　on　selecting　strategies　in　probability　to　maximum　interest.　©　2010　Springer-Verlag.