<p>Several studies argue that predicate nominals are prone to essentialize: they trigger inferences that the entities to which they are applied belong to robust and homogeneous categories with stable, explanatory membership conditions. But how consistent is this propensity? We present evidence that although predicate nominals are indeed prone to essentialize, they do so variably, and that the spectrum includes predicate nominals and predicate-nominal types whose essentializing potential is indistinguishable from that of competing non-nominal formations. The observed infra-categorical variation frequently tracks identifiable grammatical parameters, such as the semantic type of the nominal, its episodic status, and the syntactic format of the predicative phrase. We argue that this opens new opportunities for inquiry into the linguistic underpinnings of essentializing inferences.</p>