Contemporary versions of scientific realism employ explanatory inferences from the empirical success of an indispensable or working part of a theory to the truth of that part. Realism, understood this way, must therefore defend their use of such an inference against those who reject its legitimacy. I explore the merits of combining an inductive approach with an argument for the contingency of observation in order to show how the realist can provide positive reasons for believing in the legitimacy of success-to-truth inferences. Specifically, I argue that the epistemic status of success-to-truth inferences to the unobservable can be evaluated observationally.