Sixteen participants either enacted a mock crime or merely read a description of the crime. They were then given a test in which true and false three-word sentences about the crime were flashed briefly on a computer monitor while the electroencephalogram was recorded. The first two words of each sentence were always presented for 500 milliseconds, followed by a white screen for 500 milliseconds. Imperative sentence completions were then shown for 500 milliseconds. Sixteen sentences were each presented 26 times in a random order. After each sentence, the participants said either "yes" or "no." Evoked P300 responses to the sentence completions and contingent negative variations in the interstimulus foreperiod were identified. Using a bootstrap index of P300 area to compare neural responses to statements answered with deceptive "no" responses versus nondeceptive but infrequent "yes" answers, conclusive test decisions were rendered in 12 cases. There was one false positive error, five true positives, and six true negatives. The study concludes that detection that uses "event-related potentials" may develop into an effective and practical means of forensic identification. 2 figures, 1 table, and 33 references