Abstract: Current state-of-the-art virtual characters fall far short of characters produced by skilled animators in terms of behavioral adequacy. This is due in large part to the lack of emotional expressivity in physical behaviors. Our approach is to develop emotionally expressive gaze and head movement models that are driven parametrically in real-time by the instantaneous mood of an embodied conversational agent (ECA). A user study was conducted to test the perceived emotional expressivity of the facial animation sequences generated by these models. The results showed that changes in gaze and head behavior combined can be used to express changes in arousal and/or dominance level of the ECA successfully.