In this Norwegian study, bidirectional relations between children's behavior problems and child-teacher conflict and closeness were examined, and the possibility of moderation of these associations by child-care group size was tested. Eight hundred and nineteen 4-year-old children were followed up in first grade. Results revealed reciprocal effects linking child-teacher conflict and behavior problems. Effects of child-teacher closeness on later behavior problems were moderated by group size: For children in small groups only (i.e., [less than or equal to] 15 children), greater closeness predicted reduced behavior problems in first grade. In consequence, stability of behavior problems was greater in larger than in smaller groups. Results are discussed in light of regulatory mechanisms and social learning theory, with possible implications for organization of child care. (author abstract)