How are scientific phenomena discovered? How do we decide which data are erroneous and which genuine? Although these questions will find a partial answer in our experimental practices (e.g. replicability, robustness, and "calibration" of experimental results), as stressed by the philosophical movement of New Experimentalism, I am sceptical that this is the whole story that needs to be told. Rather, I believe that a good case can be made for theories sometimes providing reasons for belief in the reality of phenomena. In this project I want to argue that theories with good predictive records, high explanatory import, or even theories with virtues like unificatory power, elegance, simplicity etc. can and as a matter of fact do provide such reasons in scientific practice. This view constitutes a significant departure from the received view of the theory-evidence relationship according to which it is theories which are assessed by evidence and not vice versa.