When purchasing new graded readers for a university library, as well as making a varied collection, it is important to consider whether the books will be “popular” or “highly rated” by the students who use the library. The factors that make a book popular may be different for books that are intended to improve English skills. Therefore, using the lending history of books held by a university library, we analyze the factors that make a book popular and derive the criteria for predicting popularity in graded readers. By using these criteria, we can predict the popularity of the books we plan to purchase and so purchase the graded readers most likely to meet the needs of the library’s users. In this study, we use data extracted from an online system which stores the lending history and evaluations of graded readers in a university library, to identify the most popular graded readers and the factors related to their popularity. On the basis of these factors, we estimate an order of priority by performing statistical analysis in order to determine whether the books which the library is planning to purchase will meet the specific needs of the readers who use the library.