The Everett (many-worlds) interpretation has made great progress over the past 20-30 years, largely due to the role of decoherence in providing a solution to the preferred basis problem. This makes it a serious candidate for a realist solution to the measurement problem. A remaining objection to the Everett interpretation (and one that is often considered fatal) is that that interpretation cannot make adequate sense of quantum probabilities. Dvaid Deutsch and David Wallace have argued that, by applying decision theory to the case of a rational agent who believes in the many-worlds interpretation, we can prove that such agents _act as if_ the theory predicted objective probabilities in the sense of fundamental indeterminism, or ignorance of initial conditions. I raise the issue of whether or not this, if true, is all that the many-worlds theorist needs from \'probability\'. I first suggest a reason for thinking that the answer might be \'no\': the reason is that knowing how to act on the assumption that a given theory is true is prima facie irrelevant to the question of whether we have any reason to believe the theory in the first place. I then go on to offer a solution to this problem, drawing on resources from Bayesian confirmation theory. My conclusion is that the problem of probability in the Everett interpretation has been solved.