Research has demonstrated that people in organizations form advice networks not only for access to task knowledge, but also for the positive affect they derive from social interaction, so much so that they are willing to trade off task competence for the personal liking, pleasantness and energy experienced in interactions with colleagues. The effects of such relational choices on individual performance are poorly understood, however. I propose a theory of affect in advice networks according to which an individual’s propensity to form task-related advice ties based on positive affect experienced when interacting with a colleague enhances individual task performance. I argue that this effect changes depending on the level of activation (a feeling of being energized) generated by those interactions. Compared to forms of positive affect with neutral-to-moderate levels of activation (feelings of enjoyment), the preferential selection of work partners that elicit high-activation positive affect enhances task performance by stimulating an individual to engage in task-oriented effort. Social-network and individual performance data on 430 technology salespeople in a global IT corporation provide support for this theory. I discuss implications for network research and rational action in organizations.