Sociometric research has been concerned primarily with groups of school-age children. Yet no secondary group is more important to the development of a child's ways of interacting and, perhaps, his self-concept, than the earliest stable peer group, the nursery school. Among the relatively few sociometric studies of preschool children there has been great disparity both in method and results. The purpose of this study was, therefore, to investigate areas of disagreement and omission by means of an adapted sociometric method which incorporates both choice and rejection into a single index of status. Disagreements have arisen in every phase of sociometric research, among them the question of validation. In relation to the present investigation, an important point has been whether or not nursery school children have or can express preferences among their peers. The fact that subjects do respond has been adequate evidence for many investigators. Jennings (6) and Pepinsky (17) point out that sociometric choices, being direct measures of preferences, differ from many standardized tests, which are indirect measures of psychological processes, and thus require no validation against outside criteria. Other investigators, such as Hagman (5), Frankel (4), Moreno (11), and Emerson (3), have attempted a measure of validity by comparing verbal choices with observed contacts. Only Emerson, who used the adapted sociometric method employed in the present study, found a high degree of agreement between these variables. Both Hagman and Frankel found low correlations, but they came to differing conclusions: Hagman questioned the validity of sociometric testing on the preschool level; Frankel, on the other hand, believed that the two measures reflected two aspects of, rather than the identical, social phenomenon, pointing out that some children who had been chosen verbally were inaccessible on the playground. The same point of view was expressed by Emerson, who noted that contacts involved any observed interactions among children, including those not desired as well as those sought. Similar reasoning may explain the differences in results reported when teacher ratings of children's status were compared with the subjects' verbal choices. Koch (8) found a high positive correlation, while Lippitt (io), using the same methods, found a low degree of agreement.