Book borrowing is a key service in libraries. Library users frequently visit the library for borrowing books compared to other library services. To predict book-borrowing service in a college library, Auto Regressive Integrated Moving Average (ARIMA) model has been developed from the data pertaining to book borrowing during the year 1998 to 2013. The study found that the number of books borrowed one month and twelve months earlier could estimate the number of books borrowed in a month. The study used a fitted model for predicting book borrowing for the year 2014 by two alternative approaches: 12-steps ahead versus 1-step ahead. The calculations show that there was no significant difference (P=0.928; Wilcoxon signed rank test) between 1-step and 12-steps ahead approach for predicting book borrowing. However, the Root Mean Squared Error (RMSE) in 1-step ahead approach (109.57) was lower than 12-steps approach (131.33). The study findings indicate that ARIMA models are useful for monitoring book borrowing in institutional libraries. Furthermore, these models can predict library usage trends.