Background Several risk scores have been developed to predict acute kidney injury (AKI) after cardiac surgery. We evaluated the accuracy of eight prediction models using the gray zone approach in patients who underwent aortic surgery. Patients and Methods We retrospectively applied the risk scores of Palomba, Wijeysundera, Mehta, Thakar, Brown, Aronson, Fortescue, and Rhamanian to 375 consecutive adult patients undergoing aortic surgery with cardiopulmonary bypass. The area under the receiver operating characteristic curve (AUC) and gray zone approach were used to evaluate the accuracy of the eight models for prediction of AKI, as defined by the RIFLE criteria. Results The incidence of AKI was 29% (109/375). The AUC for predicting AKI requiring dialysis ranged from 0.66 to 0.84, excluding the score described by Brown et al (0.50). The AUC for predicting the RIFLE criteria of risk and higher ranged from 0.57 to 0.68. The application of gray zone approach resulted in more than half of the patients falling in the gray zone: 275 patients (73%) for Palomba, 221 (59%) for Wijeysundera, 292 (78%) for Mehta, 311 (83%) for Thakar, 329 (88%) for Brown, 291 (78%) for Aronson, 205 (54%) for Fortescue, and 308 (82%) for Rhamanian. Conclusion More than half of the patients in our study sample were in the gray zone of eight scoring models for AKI prediction. The two cutoffs of the gray zone can be used when using risk models. A surgery-specific and more accurate prediction model with a smaller gray zone is required for patients undergoing aortic surgery.