The antirealist argument from the underdetermination of theories by data relies on the premise that the empirical content of a theory is the only determinant of its belief-worthiness (premise NN). Several authors have claimed that the antirealist cannot endorse NN, on pain of internal inconsistency. I concede this point. Nevertheless, this refutation of the underdetermination argument fails because there are weaker substitutes for NN that will serve just as well as a premise to the argument. On the other hand, antirealists (...) have not made a convincing casefor NN (or its weaker substitutes) either. In particular, I criticize van Fraassen's recent claim that all ampliative rules in epistemology must be rejected on the grounds that they lead to incoherence. The status of the underdetermination argument remains unsettled. (shrink)