This longitudinal study examined the differences between extremely popular children and their peers, and the persistence of the patterns of differences. Teachers identified 19 extremely popular children in kindergarten through sixth grade. These students were compared with 38 randomly selected children from the teachers' classes. A multitrait-multimethod analysis demonstrated that teachers accurately chose extremely popular children. Independent observer and peer nominations, but not student self-ratings, confirmed teacher accuracy. Popular children tended to be more adept in intellectual ability, social-emotional control, social skills, and physical competence. Six years later, the 10 extremely popular students and 9 randomly selected students still in the schools were reexamined. When compared to their classmates in secondary school, the extremely popular children were still popular but did not maintain the extreme popularity they had in grade school. Some teacher ratings from grade school correlated with self-ratings. Peer nominations did not show the same strong patterns of relationships among the various personality characteristics found earlier. (Author/RJC)