In the standard view, A acts paternalistically toward B if and only if: (i) A restricts B’s liberty, (ii) A acts against B’s will, (iii) A acts for B’s own good. For example, the state may tax or prohibit smoking in the interest of citizens’ health in circumstances in which such measures are resisted by them or some of them. Telling counterexamples have been produced to each of these conditions. In the revised view, A acts paternalistically toward B if and only if: (i) A acts so as to influence B by the use of means other than rational persuasion; (ii) A does not regard B’s will as structurally decisive (i.e., A takes the prevention of voluntary self-regarding harm to constitute a reason for influencing B); (iii) A does so for B’s good or to affect matters within B’s legitimate sphere of control; (iv) A’s act cannot be justified without counting its beneficial effects on B in its favor. The wrongness of paternalism lies in the way in which a paternalistic act by A toward B infringes B’s autonomy: A does not consider B’s will authoritative in determining how A should treat B in B’s self-regarding matters―A subjects B’s will to his in this sense. Hard paternalism as thus understood should be distinguished from soft paternalism or anti-paternalism. According to the latter, the prevention of voluntary self-regarding harm is never a good reason for interference. The latter is justifiable only to prevent involuntary self-regarding harm―harm pertaining to acts that are not his or do not represent his values or preferences. Hard paternalism may, pace what soft paternalism or anti-paternalism claims, sometimes be justifiable. This is particularly so when the voluntary self-regarding harm involved is significant and the infringement of liberty required to prevent it limited or acceptable given the harm at stake. The question of when a good or an advantage is profound and when an infringement of liberty is limited is, however, difficult and worthy of further investigation. Paternalistic justifications should be distinguished from other liberty-limiting principles. That is, they should, first, be distinguished from moral paternalism focusing on improving the person’s moral character and hence his moral well-being or on making the person better (as opposed to the improvement of the person’s physical and psychological condition focused on by ordinary or welfare paternalism). Second, it should be distinguished from legal moralism concerned with barring conduct that is intrinsically morally bad (that is, bad for reasons independent of how it affects people’s character or their physical or psychological condition).