This　paper　gives　a　principal-agent　model　for　multi-agent　cooperation,　and　analyzes　the　usual　agent　cooperation　methods　with　this　model,　finally　gives　examples　and　discusses　of　this　model.　Through　the　design　of　the　incentive　contract　that　the　principal　provides　for　the　agent,　this　model　can　stimulate　the　agent　to　do　its　best　for　the　principal　in　cooperation,　and　to　accomplish　the　cooperation　of　the　principal＇s　will,　even　they　are　in　the　condition　of　asymmetric　information　about　their　cooperation.　So　this　model　can　restrain　the　agent　deception　in　cooperation.　It　also　provides　a　method　for　calculating　how　much　the　cooperation　initiator　should　pay　for　its　cooperative　partner　for　making　the　cooperation.