In section 96 of Reasons and Persons, Derek Parfit offers his now familiar tripartite distinction among candidates for ‘what matters’: (1) Relation R with its normal cause; (2) R with any reliable cause; (3) R with any cause. He defends option (3). This paper tries to show that there is important ambiguity in this distinction and in Parfit's defence of his position. There is something strange about Parfit's way of dividing up the territory: I argue that those who have followed him in viewing the choice among (1)–(3) as the (or an) important question in thinking about ‘what matters’ are mistaken, and that they bypass what seems to be a more important, even crucial, set of options and considerations. I am less concerned with what he does say than with what he ought to say, given his intuitions and arguments, and the general framework within which he is working. And I am particularly concerned to show that whether or not I am correct about what he is doing with his tripartite distinction, it is a distinction with which we should not be particularly concerned in the analysis either of what matters or of psychological continuity.