Studies on the quality of nature of science (NOS) representations in school science textbooks report them being mostly of implicit manner and not fully adequate. However, the often underlying NOS framework of the consensus list in these studies is criticized as undifferentiated and inadequate. The family resemblance approach (FRA) to NOS shows potential to give differentiated insights into the appropriateness of NOS representations with avoidance of specifying certain philosophical directions. Based on a fine-grained differentiated FRA category system (11 main categories, e.g., "knowledge"; 52 subcategories, e.g., "hypotheses"), the quality of cognitive-epistemic NOS representations identified in seven biology school textbooks from Germany was analyzed. For this, a category system was developed. Cognitive-epistemic NOS representations in four chapters of each of the seven textbooks were evaluated regarding manner (implicit, explicit) and adequacy (adequate, (partly) not adequate). Results indicate, among others, that explicit representations of the cognitive-epistemic system of science were mainly placed in the introduction chapters, whereas subject-related chapters include mostly implicit representations. In this article, we present the evaluation of the quality of cognitive-epistemic NOS representations and discuss implications for science education.