The stability of sociometric measures and their use for classifying elementary-age children are discussed. This is an important topic because of the assumption that childhood social maladjustment is a very stable and consistent phenomenon. Sociometric scores were collected over a 7-month period in a study of potential developmental differences between second and fifth grade children. Of 231 children from a middle to middle-lower socioeconomic school district outside a large Midwestern city, 53 were classified as popular children, 48 as rejected by peers, 20 as neglected by peers, 27 as controversial, and 83 as average. Over the 7-month period, sociometric groups were found to be consistent on the average of 32 percent. Rejected children were the most consistent group (52 percent), followed by popular children (43 percent), neglected children (23.5 percent), sociometrically average children (20.5 percent), and controversial children (18.5 percent). It is concluded that classification based on only one data collection may be problematic for the assignment of children to sociometric groups for research purposes, and especially so for the selection of children for clinical intervention. (RH)