Duncan Pritchard has recently defended a view he calls ‘epistemological disjunctivism’, largely inspired by John McDowell. I argue that Pritchard is right to associate the view with McDowell, and that McDowell’s ‘inference-blocking’ argument against the sceptic succeeds only if epistemological disjunctivism is accepted. However, Pritchard also recognises that epistemological disjunctivism appears to conflict with our belief that genuine and illusory experiences are indistinguishable (the ‘distinguishability problem’). Since the indistinguishability of experiences is the antecedent in the inference McDowell intends to block, I suggest that his argument rests on an inconsistent set of premises. In support of this, I show that Pritchard’s response to the distinguishability problem is incompatible with the conclusion of the ‘inference-blocking’ argument, and that the response available in McDowell’s work relies on a mistaken conception of fallibility. Either McDowell must deny the sceptic’s premise that perceptual experiences are indistinguishable, or he must give up his conclusion that perceptual warrant can be indefeasible.