The present study investigates whether nonverbal behavioral cues to hidden criminal intentions during the build-up phase of a criminal act can be measured. To this end, we created recordings of actors once in a search situation and once committing a mock crime (theft or bomb placing) in a public crowded area. For ecological validation, we used authentic CCTV footage of real crimes in Experiment I. In this experiment, the two behavioral clusters pattern of movement in space and nonverbal communication behavior were analyzed. The results showed a deviance in pattern of movement in space for offenders' compared with the nonoffenders' condition as well as a bystanders' baseline. There was no significant difference between nonverbal communication behavior in the offenders' and nonoffenders' conditions. Experiment 2 was conducted to examine the two behavior clusters use of object- and self-adaptors while controlling for interpersonal differences. The results showed an increased use of object- and decreased use of self-adaptors during the build-up phase of a mock crime compared with a control condition (search). Thus, nonverbal behavior of offenders seems to differ from nonverbal behavior of nonoffenders. However, this holds only under the conditions of a valid baseline and of judging not only a single, typical behavioral cue but a whole cluster of nonverbal behaviors, such as pattern of movement in space or use of object-adaptors in general.