【正文】 its knowledge workers, the knowledge workers might perceive that they have openended employment arrangements based upon mutual trust, thereby, are willing to perform higher level inrole behaviors and OCBs. This finding not only empirically supports Wright and Boswell’s contention that psychological contracts can be best viewed as the linking mechanism between HR systems and employee behaviors, but also provides a possible explanation as to how a mitmentbased HR system influences knowledge workers’ role behaviors. Our results also indicated that mitmentbased HR systems could positively influence knowledge workers’ relational psychological contracts and, in contrast, negatively influence their transactional psychological contracts. This finding contributes to our understanding of the influences of mitmentbased HR systems on employees’ beliefs concerning their own employment. Finally, we found that only relational psychological contracts were positively associated with inrole and extrarole behaviors. This finding not only contributes to our understanding of the impact of these contract forms on organizationally relevant outes, but is also beneficial for developing a deeper theoretical understanding of the nature of the psychological contracts. The results did not support our hypothesis that transactional psychological contracts mediated the relationship between mitmentbased HR systems and inrole behaviors. This was due to the absence of the significant relationship between transactional psychological contracts and inrole behaviors. Since relational psychological contracts were highly correlated to transactional psychological contracts and the correlation between the relational psychological contracts and inrole behaviors was higher than the correlation between transactional psychological contracts and inrole behaviors, we doubt that the relationship between transactional psychological contracts and inrole behaviors might be suppressed by relational psychological contracts. In addition, the stronger correlation between mentbased HR systems and inrole behaviors than the one between transactional psychological contracts and inrole behaviors might be another reason for the unsupported mediation effect of transactional psychological contracts. Furthermore, the failure of finding the significant mediating effect of transactional psychological contracts in the relationship between mitmentbased HR systems and inrole behaviors revealed the necessary reconsideration of the linkage among mitmentbased HR systems, transactional psychological contracts, and inrole behaviors. More specifically, organizations need different types of HR systems to form knowledge workers’ transactional psychological contracts and then to influence their inrole behaviors. For example, based on the value and uniqueness of human capital, Lepak and Snell’s HR architecture divided employees into four different types, namely, knowledgebased employment, jobbase demployment, contract work, and alliance/partnership. In order to effectively and efficiently manage these employees, Lepak and Snell further proposed four corresponding HR configurations: mitmentbased, productivitybased, pliancebased, and collaborativebased. Since our findings indicated that relational psychological contracts could link the relationship between mitmentbased HR configurations and knowledge worker’s role behaviors, transactional psychological contracts might link other types of employment relationships and HR configurations. Correspondingly, we suggest that future research explore the linking mechanism of other types of psychological contracts under the different types of employment modes and HR systems. Our crosssectional research design limits the extent to which causeeffect relations can be inferred from our findings. For example, it is intuitively plausible that organizations that adopt a mitmentbased HR system might enhance their knowledge workers’ OCBs. However, the crosssectional design implies that we measured the mitment based HR system after the OCBs period resulting in prediction of past OCBs. Future research that employs a longitudinal research design would be better suited to addressing the causal and mediated effects examine