Currently, polygraph examinations in Japan use the concealed information test (CIT) to determine whether a suspect knows specific details of a crime. The present study examined the accuracy of the CIT as a memory detection technique in a mock-theft experiment. Participants were randomly assigned to either an encoding or non-encoding group. An expert polygrapher who was not informed of the group assignments, conducted a CIT that consisted of two questions. One inquired about a card number chosen by the participant, and the other regarded an item that had been stolen. Analyses focused on the second question. Roughly 20% of cases were judged inconclusive while sensitivity and specificity for the remaining cases were 86% and 95%, respectively. Analysis was repeated using modified Lykken scoring, and rates of inconclusive cases, sensitivity, and specificity by this method were 25%, 83%, and 91%, respectively.