The main goal of this study is to provide empirical evidence for a theoretical mechanism underlying cross-informant discrepancies (CID), which occur between reports of different informants (e.g., children/adolescents and parents) of children’s/adolescents’ problem behavior. Studies comprehensively corroborate the existence of CID. However, an explanation of CID is rudimentary and inconsistent. Respective research often suffers from methodological problems and is often atheoretical. Addressing these critics, this study uses polynomial regression and is based on research on mind perception and anchoring-and-adjustment theory. It was assumed that higher CID are associated with parents’ perceived similarity to their children, whereas lower CID are related to parents’ perspective-taking efforts. Analyses were based on N = 168 parent–child dyads (children’s mean age: 12.50 years). Reports on problem behavior displayed substantial mean differences and medium-sized correlations. Polynomial regressions on CID partly supported the influence of parents’ perceived similarity and perspective taking efforts on CID. Results are discussed in the context of a possible theoretical fundament for CID.