Lewis's work on causation was governed by a familiar methodological approach: the aim was to come up with an account of causation that would recover, in as elegant a fashion as possible, all of our firm “pre-theoretic” intuitions about hypothetical cases. That methodology faces an obvious challenge, in that it is not clear why anyone not interested in the semantics of the English word “cause” should care about its results. Better to take a different approach, one which treats our intuitions about cases merely as guides in the construction of a causal concept or concepts that will serve some useful theoretical purpose. I sketch one central such purpose, suggesting, first, that an account of causation that, like Lewis's, gives a central role to counterfactuals is well-suited to fulfill it, and, second, that the most famous pre-emption-based counterexamples to a counterfactual account yield an important constraint on a successful account.