We investigated whether modal information elicited empirical effects with regard to discourse processing. That is, like tense information, one of the linguistic factors shown to be relevant in organizing a discourse representation is modality, where the mood of an utterance indicates whether or not it is asserted. Event-related potentials (ERPs) were used in order to address the question of the qualitative nature of discourse processing, as well as the time course of this process. This experiment investigated pronoun resolution in two-sentence discourses, where context sentences either contained a hypothetical or actual Noun Phrase antecedent. The other factor in this 2x2 experiment was type of continuation sentence, which included or excluded a modal auxiliary (e.g., must, should) and contained a pronoun. Intuitions suggest that hypothetical antecedents followed by pronouns asserted to exist present ungrammaticality, unlike actual antecedents followed by such pronouns. Results confirmed the grammatical intuition that the former discourse displays anomaly, unlike the latter (control) discourse. That is, at the Verb position in continuation sentences, we found frontal positivity, consistent with the family of P600 components, and not an N400 effect, which suggests that the anomalous target sentences caused a revision in discourse structure. Furthermore, sentences exhibiting modal information resulted in negative-going waveforms at other points in the continuation sentence, indicating that modality affects the overall structural complexity of discourse representation.