Purpose-To function effectively, police must separate lies from truth. Police, ideally, would be experts at this task, yet there is debate surrounding whether expertise in detecting deception is possible. Drawing upon literature outside of deception detection, the purpose of this paper is to explore whether subjects making deception judgments can improve their performance. Design/methodology/approach-The sample was 19 students from two graduate-level classes. Subjects viewed six sets of videos over eight weeks. The first five sets displayed individuals reporting whether they cheated on an exam. The sixth set displayed individuals reporting whether they had committed a (mock) robbery. After each video, subjects judged whether the videoed individual was truthful, and then the actual status was revealed. Findings-Subjects’ accuracy improved consistently over the first five sets of videos; from about 69 percent accuracy to about 89 percent accuracy. However, the accuracy for the sixth set of videos dropped dramatically to 46 percent. The results indicate that expertise in deception detection may be possible, but is likely to be limited in terms of generalizability. Research limitations/implications-The actual environment of police investigations is more variable than the study’s setting. Future studies should integrate inaccurate and incomplete feedback, which are realistic characteristics of investigations. Practical implications-The findings suggest that, like other areas of expertise, it may be possible to develop expertise in detecting deception through the use of deliberate practice with accurate feedback. Originality/value-This study combines three literature-substantiated criteria for developing deception detection expertise. The study environment involved regularity. Subjects were deliberately practicing, and subjects received accurate feedback.