The development of spatial perspective-taking has been described in a two-level model. Inconsistent research results led to more differentiated three-level models.This study draws upon these ideas and aims at developing and validating an integrated three-level model by choosing two approaches. First, prior research results wil be interpreted by using the three-level model. Second, we re-analyze data from an interview study with 95 first graders. In particular, the children’s explanations for their solutions showed that level-3-tasks were more difficult for them than level-2-tasks. Our findings suggest that a three-level model is appropriate to describe the development of spatial perspective-taking more precisely.