We respond to Conor McHugh's claim that an evaluative account of the normative relation between belief and truth is preferable to a prescriptive account. We claim that his arguments fail to establish this. We then draw a more general sceptical conclusion: we take our arguments to put pressure on any attempt to show that an evaluative account will fare better than a prescriptive account. We briefly express scepticism about whether McHugh's more recent ‘fitting attitude’ account fares better.