Acute kidney injury (AKI) is a common complication during hospitalization and is an accepted risk factor for in-hospital mortality. However, the association of severity of AKI with the long-term risk of death is not well defined. To examine the independent effect of the severity of AKI on long-term risk of death following acute myocardial infarction (AMI), we performed an observational study of 147007 elderly Medicare patients admitted for AMI from January 1994 through February 1996 as a part of the Cooperative Cardiovascular Project. We evaluated the association between AKI and all-cause mortality. We defined AKI as absolute changes in serum creatinine level, categorized as none (creatinine level increase, < or =0.2 mg/dL), mild (0.3-0.4 mg/dL increase), moderate (0.5-0.9 mg/dL increase), and severe (> or =1.0 mg/dL increase). Overall, 19.4% of the patients had AKI, including 7.1% with mild AKI, 7.1% with moderate AKI, and 5.2% with severe AKI. Less than 10% of patients who had severe AKI were alive at 10 years compared with 12.2%, 21.1%, and 31.7% patients with moderate, mild, and no AKI, respectively. The adjusted hazard ratio for death for in-hospital survivors at 10 years was 1.15 (95% confidence interval [CI], 1.12-1.18) for mild AKI, 1.23 (95% CI, 1.20-1.26) for moderate AKI, and 1.33 (95% CI, 1.28-1.38) for severe AKI. Similar results were obtained in several secondary analyses that included inpatient mortality, excluded mortality in the first 3 years, and stratified by some specified high-risk groups. Moderate or severe AKI were comparable in strength with other known correlates of cardiovascular mortality. Acute kidney injury has an independent and graded association with long-term mortality. These results should stimulate additional mechanistic and interventional studies and plans for follow-up of patients with AKI after discharge.