Intra-individual variability in kidney function is a common phenomenon; however, predictors of kidney function variability and its prognostic significance are not known. To examine this question, we assembled a cohort of 51,304 US veterans with an estimated glomerular filtration rate (eGFR) <60 ml/min at the end of the study period and who had at least two eGFR measurements during the previous 3 years. Variability in kidney function was defined for each patient as the coefficient of variation of the regression line fitted to all outpatient measures of eGFR during this time frame. In adjusted analyses, blacks, women, and those with Current Procedural Terminology and ICD-9-CM diagnostic codes for hypertension, diabetes, cardiovascular disease, peripheral artery disease, chronic lung disease, hepatitis C, dementia, acute kidney injury, and those with a greater number of hospitalizations had greater variability in eGFR. After a median follow-up of 4.9 years, there were 23.66%, 25.68%, and 31.23% deaths among patients in the lowest, intermediate, and highest tertiles of eGFR variability, respectively. Compared with the referent (those in the lowest tertile), patients in the highest tertile had a significantly increased risk of death with a hazard ratio of 1.34 (1.28-1.40), an association consistently present in all sensitivity analyses. Thus, our results demonstrate that greater variability in kidney function is independently associated with increased risk of death.