PURPOSE: To assess the predictive validity of a modified Sequential Organ Failure Assessment (SOFA) score, of which the renal component was replaced with Kidney Disease Improving Global Outcomes (KDIGO) classification of Acute Kidney Injury (AKI). MATERIALS AND METHODS: Using a prospective cohort study on AKI in Japan, we replaced the renal component of SOFA score with AKI stages according to the KDIGO criteria except that initiation of renal replacement therapy was assigned four points. We assessed the predictive validity of KDIGO-based SOFA score for hospital and ICU mortality by comparing the areas under the receiver operating characteristic curve (AUC) derived from logistic regression models with that of the original SOFA score. RESULTS: 2292 patients were registered. Overall hospital mortality was 11.6%, and ICU mortality was 5.1%. KDIGO-based SOFA score was moderately correlated with APACHE II score (rho = 0.476). The AUC for hospital and ICU mortality of KDIGO-based and the original SOFA score were 0.749 vs 0.745 (p = .393) and 0.790 vs 0.791 (p = .900). CONCLUSIONS: The prognostic performance of KDIGO-based SOFA score was not superior but comparable to that of the original SOFA score.