Consider a group of individuals cooperating in a way that results in a collectively committed wrong. Some of the individuals contribute a lot. Others contribute little. Some fail to contribute anything at all. How do we make sense of individual accountability in such cases? We might think that each individual’s accountability is limited by her causal contribution. But this sort of account is infamous for its counterintuitive implications. I present an alternative account. I make the case for thinking that distinct aspects of human agency, normally “wrapped up” in a single person, can be “distributed” practically across different people. We “distribute” agency routinely, by forming promises, by making requests, by issuing demands, and by undertaking shared action. This resulting division of agential labor makes possible a distinctive way in which one person can be accountable for the actions of another. More specifically, cooperants have the function of constitutively determining the purpose for which the others act. I analyze such a purpose by invoking Joseph Raz’s conception of a ‘protected reason’. Where the purpose that a cooperant furnishes is morally bad, she can be accountable for that wrong-making feature of what the others do. I call this phenomenon “authority-based accountability” and argue that it helps make sense of individual accountability in cooperative contexts.