In the first experiment, 41 undergraduate students at a southern university viewed videotapes of constructed testimony by 4 undergraduate students from another university. Based on confidence ratings by other students, two of the four witnesses selected were rated high on confidence, and two were rated low. Two of the witnesses were White females, and two were White males. The composed videotape depicted a male exhibiting low confidence when describing a false account, followed by a female witness with high confidence telling a true story, a female with low confidence describing a true story, and a male high in confidence describing a false story. Mock jurors were asked to assess the truthfulness of each witness. Three months following the first experiment, 217 undergraduate students in the same university participated in a second study that differed from the first in having 3 sets of instructions: no confidence instructions, don't use confidence instructions, and use confidence instructions. For both studies, researchers attempted to determine whether the mock jurors used witness confidence in determining whether a witness' statement was true. Consistent with Ekman and O'Sullivan (1991) and Landry and Brigham's (1992) findings, lie-detection ability in the untrained college students was no better than chance (50 percent). Participants were most likely to indicate they had used witness confidence when making their decision if they had been explicitly told to do so. Even then, very few reported they used witness confidence as a guide to truthfulness. The fact that the mock jurors typically rated false stories as more believable than true stories, along with the rating of high-confidence testimonies as more truthful, illustrates the difficulties novices face in assessing witness dishonesty. 2 tables and 12 references