Mirroring and synchronization of non-verbal behavior is an important characteristics of human conduct also in communication. The aims of this paper are to analyze the occurrences of mirroring gestures, which comprise head movements, facial expressions, body postures and hand gestures, in first encounters and to determine whether information about the gestures of an individual can be used to predict the presence and the class of the gestures of the interlocutor. The contribution of related speech token is also investigated. The analysis of the encounters shows that 20–30% of the head movements, facial expressions and body postures are mirrored in the corpus, while there are only few occurrences of mirrored hand gestures. The latter are therefore not included in the prediction experiments. The results of the experiments, in which various machine learning algorithms have been applied, show that information about the shape and duration of the gestures of one participant contributes to the prediction of the presence and class of the gestures of the other participant, and that adding information about the related speech tokens in some cases improves the prediction performance. These results indicate that it is useful to take mirroring into account when designing and implementing cognitive aware info-communicative devices.