Background Cardiac surgery-related acute kidney injury has short- and long-term impact on patients' risk for further morbidity and mortality. Consensus statements have yielded criteria—such as the risk, injury, failure, loss, and end-stage kidney disease (RIFLE) criteria, and the Acute Kidney Injury Network (AKIN) criteria—to define the type and consequence of acute kidney injury. We sought to estimate the ability of both the RIFLE and and AKIN criteria to predict the risk of in-hospital mortality in the setting of cardiac surgery. Methods Data were collected on 25,086 patients undergoing cardiac surgery in Northern New England from January 2001 to December 2007, excluding 339 patients on preoperative dialysis. The AKIN and RIFLE criteria were used to classify patients postoperatively, using the last preoperative and the highest postoperative serum creatinine. We compared the diagnostic properties of both criteria, and calculated the areas under the receiver operating characteristic curve. Results Acute kidney injury occurred in 30% of patients using the AKIN criteria and in 31% of patients using the RIFLE criteria. The areas under the receiver operating characteristic curve for in-hospital mortality estimated by AKIN and RIFLE criteria were 0.79 (95% confidence interval: 0.77 to 0.80) and 0.78 (95% confidence interval: 0.76 to 0.80), respectively ( p = 0.369). Conclusions The AKIN and RIFLE criteria are accurate early predictors of mortality. The high incidence of cardiac surgery postoperative acute kidney injury should prompt the use of either AKIN or RIFLE criteria to identify patients at risk and to stimulate institutional measures that target acute kidney injury as a quality improvement initiative.