Kidney function predicts cardiovascular and all-cause mortality, but little is known about the association of changes in estimated GFR (eGFR) with clinical outcomes. We investigated whether 3- and 9-yr changes in eGFR associated with risk for coronary heart disease (CHD) and all-cause mortality among 13,029 participants of the Atherosclerosis Risk in Communities (ARIC) Study. After adjustment for baseline covariates including eGFR in Cox proportional hazards models, the quartile of participants with the greatest annual decline (annual decline > or =5.65%) in eGFR were at significantly greater risk for CHD and all-cause mortality (hazard ratio 1.30 [95% confidence interval 1.11 to 1.52] and 1.22 [95% confidence interval 1.06 to 1.41], respectively) compared with the third quartile (annual decline between 0.33 and 0.47%). We observed similar results when we analyzed 9-yr changes in eGFR. Adjustment for covariates at the second eGFR used to estimate change reduced the association with CHD but not with mortality. Among participants with stage 3 chronic kidney disease, an increase in eGFR during the first 3 yr also associated with a higher risk for mortality, perhaps as a result of clinical instability. In conclusion, a steeper than average decline in eGFR associates with a higher risk for CHD and all-cause mortality. Increases in eGFR among participants with chronic kidney disease associate with similar increased risks.