In the study phase, college student participants read different passages that detailed a crime in a specific location (guilty), same location without a crime (innocent but knowledgeable), or unrelated location without a crime (naive). During electroencephalographic collection, participants viewed context phrases, some of which were relevant to the crime scene, and test words that were either congruent or incongruent to the context phrases. In the first experiment, both guilty and innocent but knowledgeable participants were originally classified as guilty because both produced large P300 amplitudes to crime-relevant stimuli. However, few innocent but knowledgeable participants produced significant N400 amplitudes and thus were rejected as guilty in the final analysis. In the second experiment, participants were assigned to either the guilty or the innocent but knowledgeable condition. Seventy-two hours after reading the appropriate passage, participants were divided into two groups, those reminded of the passage read earlier and those who were not reminded. There was no effect of reminding guilty participants on the discrimination but there was a strong effect of reminding on innocent but knowledgeable participants, suggesting a possible technique for discriminating guilty knowledge from other knowledge formats. 13 references and 14 tables