Extensive reading (ER) entails starting with simple books (graded readers) and gradually expanding the range of reading to include progressively more difficult books. In Japan, ER programmes often make use of “Yomiyasusa Level” (YL), which indicates a book’s difficulty performed by the subjectivity of Japanese people who have already done a lot of ER. One of the principles of ER is that learners should use the YL scale to choose books at an appropriate reading level. Even after learners have progressed from graded readers, they will still benefit from being able to choose books on the basis of YL, but relatively few non-graded readers appear in YL databases. The goal of this research is to estimate the YL of general English books to which a YL has not yet been assigned. This will also be useful for determining the YL scores of new graded readers and for checking the validity of previously assigned YL. First, we examine the parameters which contribute to YL using a corpus tool called Coh-Metrix, which quantitatively evaluates the linguistic characteristics of a text. Then, we perform a regression analysis using these parameters and estimate the YL of general English books.