This paper aims to investigate whether the various texts in Korean textbooks have been correctly selected according to the different levels and whether the level of various texts show syntactic coherence within the same difficulty levels. For this purpose, three sorts of texts (main text, hearing text, and reading text) in Korean published by Seoul National University were taken and analysed into syntactically-tagged corpus from which syntactic complexity of matrix clauses and embedded clauses were extracted. The syntactically-tagged corpus were classified into 12 different texts, according to the different levels and the sort of texts. After comparing those texts from various points of view, it was concluded that the level of three sorts of texts in Korean do not show syntactic coherence within the same textbooks. In addition, only the main texts has been correctly selected according to the different levels. Following these results, it is suggested that especially the hearing texts in Korean 3 be made more difficult by modifying embedded clauses in those texts.