This thesis examines the role of emotion in face recognition, using measures of the visual scanpath as indicators of recognition. There are two key influences of emotion in face recognition: the emotional expression displayed upon a face, and the emotional feelings evoked within a perceiver in response to a familiar person. An initial set of studies examined these processes in healthy participants. First, positive emotional expressions were found to facilitate the processing of famous faces, and negative expressions facilitated the processing of novel faces. A second set of studies examined the role of emotional feelings in recognition. Positive feelings towards a face were also found to facilitate processing, in both an experimental study using newly learned faces and in the recognition of famous faces. A third set of studies using healthy participants examined the relative influences of emotional expression and emotional feelings in face recognition. For newly learned faces, positive expressions and positive feelings had a similar influence in recognition, with no presiding role of either dimension. However, emotional feelings had an influence over and above that of expression in the recognition of famous faces. A final study examined whether emotional valence could influence covert recognition in developmental prosopagnosia, and results suggested the patients process faces according to emotional valence rather than familiarity per se. Specifically, processing was facilitated for studied-positive faces compared to studied-neutral and novel faces, but impeded for studied-negative faces. This pattern of findings extends existing reports of a positive-facilitation effect in face recognition, and suggests there may be a closer relationship between facial familiarity and emotional valence than previously envisaged. The implications of these findings are discussed in relation to models of normal face recognition and theories of covert recognition in prosopagnosia.