Counterfactuals describe imagined alternatives to reality that people know to be false. Successful counterfactual comprehension therefore requires people to keep in mind both an imagined hypothetical world and the presupposed real world. Counterfactual transparency, that is, the degree to which a context makes it easy to determine counterfactuality, might affect semantic processing. This might especially be the case for languages like Chinese which lack dedicated counterfactual markers and therefore are more context-dependent. Using event-related potentials, this study investigates the role of counterfactual transparency on the comprehension of Chinese counterfactuals. For transparent contexts (e.g., “If everything in the world could go back in time . . .”), in which the information needed to identify counterfactuality is highly accessible, discourse incongruent words elicited P600 effects. In contrast, for nontransparent contexts (e.g., “If better preparations were made at that time . . .”) in which readers must attend to specific discourse context and engage pragmatic information to arrive at the counterfactual interpretation, discourse incongruencies gave rise to N400 effects. These findings suggest that (a) provided a constraining context, semantic processing is not disrupted by the dual nature of counterfactuality (i.e., readers can rapidly make contextually appropriate inferences to interpret subsequent narratives) and (b) the degree of transparency of the counterfactual can affect the nature of subsequent semantic processing. Our findings support the usage-based view that Chinese counterfactual comprehension is highly context-dependent and pragmatics-driven.