The view that, as children get older, there is a decline in the use of feature-based forms of face processing to more configurational forms of processing was examined by asking 6-year-old and 9-year-old children to judge which of two photographs matches an identical probe photograph. The probe and test stimuli were either photographs of whole faces or photographs of isolated facial features. Within this standard method, the stimuli also systematically varied in terms of the familiarity of the faces shown and in the orientation of presentation, both factors that have been interpreted as effecting configurational encoding. A number of age-related effects are observed: (a) older children are better at recognizing whole faces than younger children, (b) older children exhibit a clear face inversion effect with whole faces while the younger children are equally adept at identifying upright and inverted whole faces, and (c) analysis of the recognition rates associated with the individual features reveals that younger children are better than older children when asked to recognize eye regions. It is argued that the data support the view that as children get older there is a change in the forms of piecemeal encoding employed and an increase in configurational processing.