We show that is hard to find regular or even ordered (also known as Davis-Putnam) Resolution proofs, extending the breakthrough result for general Resolution from Atserias and Müller to these restricted forms. Namely, regular and ordered Resolution are automatable if and only if P = NP. Specifically, for a CNF formula $F$ the problem of distinguishing between the existence of a polynomial-size ordered Resolution refutation of $F$ and an at least exponential-size general Resolution proof being required to refute $F$ is NP-complete.