Twenty-two right-handed subjects were asked either to identify sad, neutral, or laughing faces presented on a computer monitor or to view passively the same pictures without classification. Visual evoked potentials were recorded from F3/4, C3/4, P3/4, O1/2, and T5/6 derivations. In comparison with the passive viewing, the emotion recognition was characterized by the higher level of cortical activity reflected in the higher N1, N2, N3 amplitudes and shortened latencies of N1, P2, and N2 waves. In contrast, the latencies of the later P3 and N3 waves were longer. During emotion recognition, the dynamical brain mapping technique revealed symmetrical activation of the frontocentral areas and only the right-side activation during the passive perception. Factor analysis demonstrated a complication of N2 structure in the task of face emotion recognition. A principal component corresponding to the descending part of the N2 wave was revealed, which probably reflected the stage of image classification.