One way of objecting to possible worlds is that we should only believe in things which are part of the causal order. Such objections are brought by Stalnaker 1984, p. 49, and McGinn 1981, pp. 153–158. But what is causation? Suppose we take the analysis in Lewis 1973b. For Lewis, causation depends on a prior notion of counterfactual dependence, where a proposition (set of worlds) b counterfactually depends on a proposition a in a world w iff a and b are true in w, and there is a world in which a and b are both false, which is closer to w than any world in which a is false and b is true. So if that notion is involved in causation, the very description of a world in causal terms brings in reference to other worlds. On the other hand, if these other worlds are not required for ontology, then causation also is not required for ontology, and the claim that other worlds and their contents are not real because they are not involved in causation collapses. One form of the causation objection to possible worlds seems to assume that causation is a relation between things, and physical things at that – so that if you claim that worlds are abstract, you may want to deny their involvement in causation on the ground that abstract entities cannot enter into causal relations.