The present study explores how well teacher trainees can detect liars. Moreover, a new method was applied to investigate beliefs that teacher trainees hold about liars. The results indicate that, overall, teacher trainees were not better than chance in detecting true and invented stories. Generally, participants reported to have used only a few cues for their credibility judgment, where most of these self-reported cues are stereotypical and invalid deception cues (e.g., gaze aversion). Further analyses with a Brunswikian lens model showed that the self-reported cues were good predictors of their credibility judgment but only poorly predictive for the objective truth/lie status of the statement. Practical implications of the results are discussed.