BACKGROUND: Cystatin C is an alternative measure of kidney function that may have prognostic importance among elderly persons who do not meet standard criteria for chronic kidney disease (estimated glomerular filtration rate [GFR] > or =60 mL/min per 1.73 m2). OBJECTIVE: To evaluate cystatin C as a prognostic biomarker for death, cardiovascular disease, and incident chronic kidney disease among elderly persons without chronic kidney disease. DESIGN: Cohort study. SETTING: The Cardiovascular Health Study, a population-based cohort recruited from 4 communities in the United States. PARTICIPANTS: 4663 elderly persons. MEASUREMENTS: Measures of kidney function were creatinine-based estimated GFR by using the Modification of Diet in Renal Disease equation and cystatin C concentration. Outcomes were death, cardiovascular death, noncardiovascular death, heart failure, stroke, myocardial infarction, and incident chronic kidney disease during follow-up (median, 9.3 years). RESULTS: At baseline, 78% of participants did not have chronic kidney disease (estimated GFR > or =60 mL/min per 1.73 m2) and mean cystatin C concentration, creatinine concentration, and estimated GFR were 1.0 mg/L, 79.6 micromol/L (0.9 mg/dL), and 83 mL/min per 1.73 m2, respectively. Cystatin C concentrations (per SD, 0.18 mg/L) had strong associations with death (hazard ratio, 1.33 [95% CI, 1.25 to 1.40]), cardiovascular death (hazard ratio, 1.42 [CI, 1.30 to 1.54]), noncardiovascular death (hazard ratio, 1.26 [CI, 1.17 to 1.36]), incident heart failure (hazard ratio, 1.28 [CI, 1.17 to 1.40]), stroke (hazard ratio, 1.22 [CI, 1.08 to 1.38]), and myocardial infarction (hazard ratio, 1.20 [CI, 1.06 to 1.36]) among these participants. Serum creatinine concentrations had much weaker associations with each outcome and only predicted cardiovascular death. Participants without chronic kidney disease who had elevated cystatin C concentrations (> or =1.0 mg/L) had a 4-fold risk for progressing to chronic kidney disease after 4 years of follow-up compared with those with cystatin C concentrations less than 1.0 mg/L. LIMITATIONS: Because this study did not directly measure GFR or albuminuria, the extent to which cystatin C may be influenced by nonrenal factors was not determined and participants with albuminuria might have been misclassified as having no kidney disease. CONCLUSIONS: Among elderly persons without chronic kidney disease, cystatin C is a prognostic biomarker of risk for death, cardiovascular disease, and chronic kidney disease. In this setting, cystatin C seems to identify a "preclinical" state of kidney dysfunction that is not detected with serum creatinine or estimated GFR.