Recently there has been a tendency on the part of some scientific realists to weaken their philosophical theses with respect to the success of science. Some of them have suggested that a satisfactorily realist standpoint should be a highly modest approach to scientific success , leaving many with the impression that scientific realism nowadays is nothing that we once thought it was. In light of that, the main concern of this paper is methodological, here I want to answer the question how far can we push the boundaries of our realist commitments and still be in control of our philosophical claims. In particular, I deal with the issue of how a certain type of weak version of selective realism will necessarily allow for true contradictions, dialetheias –even if that is not desirable. Here I argue that if one presents a very weak characterization of selective realism, one that is in line with contemporary projects, this type of realism will not forbid the possibility of things such as dialetheias. I also claim that, if that is the case, we face the following dilemma: or our general characterization of selective realism is mistaken or selective realists cannot provide a satisfactory explanation of why and how to forbid dialetheias in science