Behavioral regulation supports children to control their cognitive and emotional skills and participate fully in classroom learning and interactions. Teacher ratings are frequently used to collect data but are highly susceptible to the teacher's response bias, meaning much child level variance is attributable to the teacher level. Multilevel modeling techniques are commonly used to assess the psychometric properties of measures at different levels by disaggregating within- and between-teacher variation in student behaviors. In this study, we attempt to identify the factor structure at each level using teacher ratings on the Child Behavior Rating Scale (CBRS). The sample of 1,375 preschool children (49.2% males) was drawn from a longitudinal study in Singapore. The teacher ratings of these children were collected at three time points (average age at Time 1 was 58.5 months, Time 2 was 69.8 months, and Time 3 was 80.9 months). Using multilevel exploratory and confirmatory factor analyses, a three-factor model at the within level and a two-factor model at the between level best fit the data. At the within level, the three factors were identified as basic interpersonal skills, advanced interpersonal skills, and learning-related social skills. The three factors showed good internal consistency and differential correlates. The two factors at the between level may be related to teacher and classroom characteristics. This study provides empirical support for (a) differential factor structures of CBRS at the within and between levels; and (b) two distinct interpersonal skills. Future studies should aim at exploring the sources of between level variability. (PsycInfo Database Record (c) 2021 APA, all rights reserved).