The main aims of this study were to further validate the Children's Social Understanding Scale (CSUS), which is a parent-report measure developed by Tahiroglu and colleagues, and to fill in some gaps in the existing research. Our study included more than 700 Polish parents from a diverse educational background who had children with disabilities, including mild intellectual disability (MID) and hearing impairment (HI), or typically developing children. After conducting a confirmatory factor analysis, we built a model with acceptable fit. The model turned out to be invariant in each of our subsamples. Children without disabilities had the highest scores on the CSUS, whereas children with MID had the lowest scores. We were able to obtain not only good reliability on the measure but also a high test-retest correlation between longitudinal results over a period of 20 months. We validated the CSUS using laboratory theory of mind (ToM) tasks, as well as a social skills measure, after controlling for the language skills and age of the children.