隨著科技進步、跨文化溝通頻繁等等，領導者與追隨者越來越容易面對矛盾的工作要求，為了要能夠同時整合看似對立的需求，矛盾的能力是不可或缺的。由於至今尚未有研究釐清矛盾追隨行為在矛盾領導行為與部屬的工作績效之間，究竟是扮演著調節亦或者是中介的角色，這樣的研究缺口便引發了本研究的動機。 本研究的目的在於探索矛盾領導行為、矛盾追隨行為、任務性績效、脈絡性績效以及適應性績效間，有可能的替代關係。根據社會學習理論，追隨者的行為會受到領導者的激發，這顯示了矛盾追隨行為在矛盾領導行為與追隨者不同工作績效之間可能扮演著中介的角色，然而，從另外一方面來看，領導行為對追隨者績效的影響，有可能隨著主管部屬契合度好壞程度而有所不同，因此矛盾追隨行為也有可能是矛盾領導行為與追隨者不同工作績效的調節角色。 本研究在台灣蒐集的跨組織有效473份主管部屬配對問卷，發現矛盾追隨行為在矛盾領導行為與部屬不同工作績效間，呈現完全正向中介的效果，而非調節效果。 ;Along with the progress of technology, frequent cross-cultural communication and so on, leaders and followers are likely to face contradictory demand in organization. In order to meet competing demand in workplace simultaneously, paradoxical ability (both/and) is essential. Due to the lack of research on the subject of, if paradoxical followership behavior plays a mediator or moderator during the process of paradoxical leadership behavior affecting employees′ performance in the domains of human resource management, motivation of this study initiated. The purpose of this study was to explore alternative relationships between paradoxical leadership behavior, paradoxical followership behavior and task performance, contextual performance and adaptive performance. According to Social Cognitive Theory, the behavior of followers would be stimulated by leaders in organization. It showed the relationship between paradoxical leadership behavior and different facets of work performance might be mediated by paradoxical followership behavior. On the other hand, the influence of leadership behavior to employees′ performance may affected by the quality of person-supervisor (PS) fit. Therefore, paradoxical followership behavior might moderate the relationship between paradoxical leadership behavior and different facets of work performance. Drawing on an effective cross-organizational data of 473 supervisor-subordinate dyads from Taiwan, we found that paradoxical followership only play positively significant mediator instead of moderator in relationship between paradoxical leadership and three kinds of employees′ performance.