In a two-stage system with the two stages connected in series，fairly setting the target outputs for the first stage，which are also the inputs for the second stage，is critical for ensuring that the two stages have in-centives to collaborate with each other so as to achieve the best performance． Data envelopment analysis( DEA) ，which is a non-parameter approach for performance evaluation of decision making units ( DMUs) ， has drawn a lot of attention from many scholars． In this paper，a new two-stage DEA model taking account of fairness is proposed to set the target intermediate products in a two-stage system． We find that the optimal re-sults of this model are exactly the Nash solution of a Nash bargaining game．