Abstract: A computer-implemented method and a system for optimizing diluent recovery of a diluent recovery unit (DRU) used to recover a diluent from a tailings generated by a bitumen froth treatment process (BFTP). A regression model is determined from data points for operating conditions and corresponding diluent recovery, generated during operation of the DRU. The regression model is used to predict diluent recovery under a particular operating condition and determine a recommended value of the operating condition to achieve a target diluent recovery. The system may graphically display the regression model, the predicted diluent recovery and the recommended value, or cause the DRU to vary the operating conditions towards the recommended value.