Background: To build a predictive model for patients with dialysis-dependent acute kidney injury (AKI-D) after cardiac surgery with the cardiopulmonary bypass (CPB), according to disease severity. Methods: A single-centre, retrospective cohort study was performed to determine the demographic and clinical parameters (including the specific factor, CPB duration) for risk of poor outcome in patients requiring RRT after cardiac surgery with CPB. A new model was built for mortality prediction in these patients on the basis of the identified risk factors and Sequential Organ Failure Assessment score. Results: The newly developed model showed good discriminatory ability for predicting death in patients with AKI-D after cardiac surgery with CPB. The area under the receiver-operating characteristic (ROC) curve for the score was 0.892 (95% confidence interval, 0.852-0.925). We also determined the criterion for the choice of RRT modality by applying this model. On applying the new model in intermittent haemodialysis patients, a score of ≤3.2 was found safe for selecting the RRT modality. Conclusions: The new scoring system was valid and accurate in predicting death for AKI-D patients after open-heart surgery. This system and value for choice of RRT were determined for guidance only, to facilitate decision-making in difficult situations.