Evidence suggests that a significant proportion of school children consider themselves lonely and isolated from peers. However, less data exists as to (1) the accuracy of children's perceived loneliness when compared with their reported popularity with peers; (2) specific behavioral attributes that might contribute to popularity; and (3) stability of popularity over a variety of school-related social situations. The present experiment measured the perceived loneliness, behavioral attributes, and actual popularity of 364 boys and girls in grades 2, 4, and 6. Students were recruited from two rural/suburban, middle class and racially mixed school districts in the Midwest. Materials consisted of a loneliness scale, in which students indicated the veracity of positive and negative statements as applied to themselves, and a sociometric peer-nominating form in which students nominated three classmates for each of 14 attributes. Children were tested in class groups in their regular classrooms. Developmental differences were found for both popularity and perceived loneliness. Both developmental and gender differences were found for the importance of such variables as sense of humor, moodiness/sadness, creativity, and athletic and academic ability. Popularity was more stable over social situations (i.e., classroom projects and athletic activities) for girls than for boys. Results were discussed in light of a theory of social metacognition. (Author/CB)