With respect to group decision making problems (GDM) with preference relations in form of incomplete linguistic judgment matrices, a group ranking method is proposed based on the relative entropy. Firstly, an incomplete degree is defined which is used to distinguish the degree of acceptance with incomplete linguistic judgment matrices. Secondly, the incomplete linguistic preference is transformed into incomplete numerical preference by using the transformation formula, and an optimize model for alternative ranking vectors is constructed according to the theory of relative entropy and additive consistency. Thirdly, a proximity entropy and similarity entropy indicator is defined, and decision makers’ weights are adjusted dynamically according to the two indicators until the steady weights of decision makers are obtained, so that the final group ranking vectors are calculated. Finally, numerical examples are illustrated to verify the effectiveness of the proposed method.