Abstract: Humans reciprocate. We want to return favors we have received, but also respond appropriately to behavior that we regard as unfair against us. Whereas previous research has typically tried to isolate the most prominent explanations for reciprocal behavior - inherent preferences for reciprocity and repeated interaction - the present paper addresses the question if and how those interact. Developing a theoretical model of a long-term employment relationship, we first show that reciprocal preferences are more important when an employee is close to retirement. At earlier stages, repeated interaction is more important because more future rents (which increase players' commitment in this case) can be used to provide incentives. Preferences for reciprocity still affect the structure of an employment relationship early on, though, because of two reasons. First, preferences for reciprocity effectively reduce the employee's effort costs. Second, they allow to relax the enforceability constraint that determines the principal's commitment in the repeated interaction. We test our main predictions using data from the German Socio-Economic Panel (G-SOEP) and find cross-sectional evidence for a stronger positive effect of positive reciprocity on effort and wages for older workers.