Detrimental effects of interruptions have been widely reported in the literature, particularly with laboratory-based studies. However, recent field-based studies suggest interruptions can be beneficial, even vital to maintaining or enhancing system performance. The literature seems to be at critical juncture; how do practitioners reconcile these perspectives? Do we ban interruptions or let them flow freely? To address this, we study how interruptions affect work performance over differing units of analysis (a dyad versus an individual) in a mi-croworld scheduling task with 72 participants and a computer agent. We found that a team performance perspective shows more benefits from interruptions than an individual performance perspective. In other words, teams suffered less from the adverse effects of interruptions than individuals. Results show that systems-level aspects of interruptions, for both the individual and the team, plays a role in determining whether interruptions have a positive or negative effect.