The central thesis of this paper is that Epicurus held that swerves of the constituent atoms of agents' minds cause the agents' volitions from the bottom up. De Rerum Natura 2.216-93 is examined at length, and Lucretius is found to be making the following claims: both atoms and macroscopic bodies sometimes swerve as they fall, but so minimally that they are undetectable. Swerves are oblique deviations, not right-angled turns. Swerves must be posited to account both for cosmogonic collisions quite generally and for every "free volition," including those of animals. All volitions are fresh starts of macroscopic motion, caused by that "something in our chests" which later philosophers would call 'the faculty of will.' Since nothing can come to be from nothing, volitions must be caused from the bottom up by swerves, fresh starts in the mind's atoms motions caused by the atoms' inherent swerviness. This is what Lucretius is saying, and what Epicurus had to say in order to defend both libertarianism and atomism. Modern scholars are wrong, then, in rejecting the interpretation of Guissani and Bailey, which was crudely stated, but substantively correct. The rival interpretations of Furley, Fowler, and Englert do not do justice to Epicurus' libertarianism, and that of Sedley does not do justice to his atomism, which entails universal bottom-up causation. Epicurus did not himself draw much attention to his positive doctrine of the swerve, preferring to emphasize the untenability of the deterministic alternative. The notoriety of the doctrine in Cicero's day is due primarily to Chrysippus, who insisted that swerves cannot occur, since they would be 'uncaused' motions, and secondarily to Carneades and Zeno of Sidon.