The scientific realism debate addresses the question whether scientific theories should be taken to be approximately true or rather to constitute fictions or tools deployed for structuring the observable phenomena. String theory offers mixed messages for this debate. On the one hand, the theory's structural uniqueness and its indications of limitations to scientific underdetermination seem to be at variance with the anti-realist interpretation of theories as a tools or fictions since that interpretation would suggest freedom of choice in creating and adjusting the theories. On the other hand, some technical characteristics of string physics, in particular the concept of string duality, arguably contradict ontological realism about scientific objects. Both points may be accommodated by a position akin to structural realism, where the theory's structural aspect but not its otological import is understood realistically. Structural uniqueness and the devaluation of scientific underdetermination could be advantageous in this context, since the lack of alternatives to the string theoretical structure could provide the basis for a more distinctly realist position than classical structural realism. (see [2].)