The estimation of baseflow is a crucial task for water resources management. This research aimed to calibrate, validate, and evaluate the performance of the digital graphical method to separate baseflow. Two outlets of Brantas watersheds located at Kertosono (±6414,2 km2) and Ploso (± 8844,2 km2) were used for this study. This study used three digital graphics filters known as a local-minimum, fixed-interval and sliding-interval. The methodology consists of (1) data inventory, (2) data processing, (3) calibration, (4) validation, and (5) evaluating of models performance. Daily discharge data were separated into two periods, from 1996 to 2005 for the calibration and from 2006 to 2015 for the validation. Firstly, each method was tested annually by entering parameter values through trial and error. The period from July to September is considered as the peak of dry seasons and used to calibrate. The optimal values of parameters for calibration obtained by Averaging yearly values. Secondly, the optimal values from calibration are then used to test the model during the validation. Furthermore, statistical analysis was used to compare model performance during calibration and validation. Results show that the three methods could be used and perform well; however, the best performance is a local-minimum method.Keywords: Baseflow, separation, graphical-method, digital-filter, Brantas