In this paper, we use readers’ borrowing history records as the source data of mining. Each borrowing history record contains a reader ever borrowed books, and use clusters to find the most adaptive recommendations of boos from two aspects. One is to let one reader as the target of mining and assign his borrowing history record as the center of cluster. Then, we propose a clustering method to let each other borrowing history record is grouped with the center to which it contains the reader’s borrowing history record for satisfying the threshold of the minimum borrowing similarity. We can find the most adaptive book recommendations for the reader according to the characteristics of borrowing tendency of the cluster. The other is to let one book as the target of mining and assign it as the center of cluster. Then, we propose a clustering method to let each other borrowing history record is grouped with the center to which it contains the book. We compute the association factors of the center in the cluster, and find the most adaptive readers of borrowing the book according to the borrowing similarity between the association factors and borrowing history records. We design and construct a mining system for fining the most adaptive recommendations of books according to we propose the both methods. The results of the mining can provide very useful information to plan the services of the most adaptive book recommendations for libraries.