Research on age-related cognitive change traditionally focuses on either development or aging, where development ends with adulthood and aging begins around 55 years. This approach ignores age-related changes during the 35 years in-between, implying that this period is uninformative. Here we investigated face recognition as an ability that may mature late relative to other abilities. Using data from over 60,000 participants, we traced the ability to learn new faces from pre-adolescence through middle age. In three separate experiments, we show that face learning ability improves until just after age 30--even though other putatively related abilities (inverted face recognition and name recognition) stop showing age-related improvements years earlier. Our data provide the first behavioral evidence for late maturation of face processing and the dissociation of face recognition from other abilities over time demonstrates that studies on adult age development can provide insight into the organization and development of cognitive systems. (Contains 1 table and 6 figures.)