I consider this question in light of Slote’s proposal for an ‘agent-based’ account of social justice, on which we can evaluate a society in terms of the morally virtuous motives being expressed in the ‘actions’ of laws, institutions and customs. I first motivate Slote’s project, then present two interpretations of his account: one which assesses the actual or apparently manifest motives of a sufficient number of individuals who instantiate social and political institutions, and one where social and political institutions are thought of as discrete agents which themselves express or reflect quasi-motives , aside from any particular individual’s actual or apparent motive. I argue that both of these formulations of the agent-based picture of social justice meet the same problem. For example, with regard to individuals’ motives, what would count as the good or virtuous motive qua political participant or representative can only be properly specified by prior reference to the expectations attached to particular social or political roles. Having exposed the problems in the agent-based approach, I end by considering whether and to what extent it generalizes and threatens to hamper any kind of virtue ethical project in social and political philosophy.