Nowadays, a lot of statistical techniques for determining dynamic models aiming at defining and controlling most appropriate variables of a system has been used. One of the most used is transfer function model (ARIMAX) discussed by Box and Tiao. In this paper Transfer function technique was applied to data representing agricultural exports and exchange rate in the Sudan for the period (1956 – 2018). Augmented Dickey-Fuller (ADF) tests confirmed both series level are non stationary however, their first difference is stationary. Both ADF as well as ACF test confirmed that ARIMA(1,1,0) is appropriate model for modeling both agricultural exports and exchange rate in the Sudan. According to the application of transfer function approach proposed by Box and Tiao as well as models selection criteria, ARIMAX - TF Model (3, 0, 1) model shown smallest values of models selection criteria. Hence it is chosen as an appropriate and parsimonious transfer function model for forecasting agricultural exports data in the Sudan.