Modern philosophical debate on explanation began with Hempel and Oppenheim's 1948 article: Studies in the Logic of Explanation . Hempel and Oppenheim's view: their Deductive-Nomological model of explanation for deterministic phenomena, and the Inductive-Statistical model of explanation for indeterministic phenomena, were for many years the received view on the subject. That view has since fallen out of favour, however, and in its wake, many alternative models have been proposed. Of these, Peter Railton's Deductive-Nomological-Probabilistic model represents an exceptionally promising proposal. In this thesis, I argue that while we should accept the essentials of Railton's account, we should not endorse it in its totality. Railton's account of explanation forces us to ascribe to a hard form of reductionism. It also forces us to accept a propensity interpretation of probability and a causal interpretation of irreducibly indeterministic phenomena. In this thesis I argue that one should not have to engage in these metaphysical debates--that an account of explanation should be neutral with respect to these issues.