To determine whether teachers' dominative and integrative contacts with children are related to their assessments of children, four kindergarten teachers were observed for 300 minutes each. The Clifton form of the Anderson observational system was used to measure teacher contacts with individual children in their classes. Teachers were then asked to assess each child's likelihood of school success, family interest and socioeconomic status. Chi-square tests of goodness of fit showed that the teachers differentiated among children in their contacts. Canonical correlation analysis indicated an overall relationship between teacher contacts and teacher assessments. Children assessed high on likelihood of school success and family interest tended to receive more integrative teacher contacts. (Author/SB)