Objectives: To create a simplified clinical score to predict mortality within 28 day of severe AKI patients who receiving RRT in ICU. Methods: A prospective multicenter cohort, involving the adult patients who admitted to ICU with severe AKI in 14 centers across Thailand during 2019 to 2021. Apart from descriptive analysis, multivariable logistic regression was used to perform estimated coefficients of predictive models. A predicting score was derived from the regression equation with Receiver-Operating Characteristic (ROC) analysis for evaluating diagnostic test and predictive models. Internal validation was obtained with bootstrapping method. Results: 761 (65 %) from all 1,167 cases had received RRT. Among those of RRT, 372 (49 %) cases ended up with death. The simplified point of each mortality predictor including underlying malignancy =2, low Glasgow coma score = 2, main cause of AKI (ischemic =3, multifactorial =3, sepsis = 7), liver disease contributing AKI = 2, any complication of RRT= 4, urine volume (oliguria =1, anuria = 2). The optimism-corrected performance (area under ROC curve) of the score was 0.72 (0.69, 0.76). At the lower cutoff value of 5; the sensitivity, specificity, and negative predictive value were 0.97, 0.21, and 0.89, respectively. And at the upper cutoff value of 10; the sensitivity, specificity, and positive predictive value were 0.57, 0.77, and 0.70, respectively. Conclusions: The RRT-MORTALITY score can be considered as a clinical score to predict death outcome in severe AKI patients who receiving RRT. The simplicity and feasibility of the score will increase the possibility of use especially in resource limited settings. However, external validation must be needed before wide use.