University libraries play an important role in supporting the learning and research processes. However, book borrowing transaction data is often not optimally utilized to understand user needs. This study aims to identify book borrowing patterns at the University of Timor Library in order to find associations between book types that can be used for reading recommendations and collection management. The method used is association rule mining with the Apriori algorithm on book borrowing transaction data for the 2023–2024 period. The analysis results show strong association patterns between several book titles, particularly in the fields of finance and taxation, such as “Financial Statement Analysis and Its Application,” “Value Added Tax,” and "Techniques for Drafting Local Regulations: Regarding Local Taxes and Local Levies," which had a support value of 0.011494, confidence of 100%, and lift of 87. Another strong pattern was found in the field of information technology, namely between the books “Data Mining: [For Data Classification and Clustering]” and “A Brief Introduction to Python 3 Programming” with a lift of 58, indicating a correlation between the topics of programming and data analytics. These findings confirm that library users tend to borrow books with related themes, so the results of this study can be used for collection planning, shelf arrangement, and the development of data-based recommendation systems in libraries.