Aristotle’s requirement that virtuous actions be chosen for themselves is typically interpreted, in Kantian terms, as taking virtuous action to have intrinsic rather than consequentialist value. This raises problems about how to reconcile Aristotle’s requirement with (a) the fact that virtuous actions typically aim at ends beyond themselves (usually benefits to others); and (b) Aristotle’s apparent requirement that everything (including virtuous action) be chosen for the sake of eudaimonia. I offer an alternative interpretation, based on Aristotle’s account of loving a (...) friend for herself, according to which choosing a virtuous action for itself involves choosing it on account of those features of it that make it the kind of action it is, where these features include its intended consequences (such as the benefits it seeks to provide to others). I then suggest that Aristotle may take these consequences (including benefits to others) as contributing (and contributing non-instrumentally) to the agent’s own eudaimonia, and that there is no conflict here with Aristotle’s view that eudaimonia is an activity of the soul. For just as my activity of teaching is actualized in my students (provided they learn from me), so too my virtuous activity can be actualized in its beneficiaries. If this is right, then Aristotle’s view is far from the Stoic (and proto-Kantian) view often attributed to him. (shrink)