In this chapter we have discussed the field of facial expression analysis from both a Behavioural Science and a Computer Vision perspective. We introduced the field of facial expression analysis with Darwin's initial findings and then went on to provide details of the research conduced by Ekman et al. on facial expression analysis highlighting the importance of facial expression dynamics. We then provided details of the current state-of-the-art in automated facial expression analysis, and presented our contribution to this field. In our facial expression classification section we demonstrated the success of our PCA based technique at classifying the primary facial expressions achieving and average AUC of 0.91. We developed separate LLE based shape models for the classification of upper and lower face AUs. Context independent classifiers were used to discriminate between two of the three AUs that occur within the eyebrow area. Given our approaches to classification of static expressions, we then extended on this work to create dynamical models which estimate the AU intensity. The performance of this approach was evaluated using both the full FACS intensity system and a simpler system of low, medium, and high intensities. Distributions of the resulting intensity estimations for a sample of the Cohn-Kanade database were presented. In the final section of this chapter we described a technique which allows for photo-realistic expression synthesis (Ghent, 2005a; Ghent, 2005b). This was achieved by applying machine learning techniques to the modelling of universal facial expression mapping functions. Three mapping functions were developed for mapping from neutral to joy, surprise, and sadness. We also demonstrated how the representation of expression used allowed the intensity of the output expression to be varied. This ability to vary the intensity of output enabled us to generate image sequences of expression formation.