In factive clausal embedding ([He knows [that it is warm outside]]), the embedded clause is presupposed to be true. In non-factive embedding ([He thinks [that it is warm outside]]) there is no presupposition, and in counterfactive embedding ([It only seems [that it is warm outside]]) the embedded clause is presupposed to be false. These constructions have been investigated as a window into the complexity of language and thought, and there are disputes as to the relative contributions of lexical, syntactic or non-verbal resources in their interpretation. We designed a sentence-picture matching task to test comprehension of these constructions in a group of aphasic participants and in non-brain-damaged controls. In particular, we tested the capacity to reach a factive or counterfactive interpretation. In factive interpretation trials, participants with aphasia performed nearly as well as controls, while in counterfactive interpretation trials they performed significantly worse. Accuracy in factive and counterfactive interpretation trials correlated with other syntactic and lexical measures. Only performances on counterfactive trials correlated with non-verbal reasoning measures. Exploratory regression models suggest that verbal and non-verbal scores were separate factors. Results indicate that a disruption of counterfactive interpretation in aphasia is linked to reduction of syntactic and/or conceptual-propositional capacities.