The debate between Simulation-Theory (ST) and Theory-Theory (TT) provides the dominant theoretical framework for research on "theory of mind" (ToM). Behavioural research has failed to provide clear methods for discriminating between these theories, but a number of recent studies have claimed that neuroimaging methods do allow key predictions of ST and TT to be tested. In the current paper it is argued that neuroimaging studies have not in fact provided any data that discriminates between ST and TT accounts of propositional attitude ascription, and moreover that it is uncertain that they will in the future. However, it is also argued that the fault lies with the ST/TT debate, not with the methods and concepts of neuroimaging research. Neuroimaging can certainly contribute to our understanding of ToM, and should contribute to the project of developing theoretical models more firmly grounded in specific cognitive and neural processes than ST or TT.