OBJECTIVES: The objective of this research was to estimate expenditures attributable to chronic kidney disease (CKD) by disease stage and to identify predictors of variation in spending.  METHODS: A retrospective analysis of de-identified health insurance claims, laboratory data and pharmacy claims was performed using the MarketScan Lab Results database for 2007.   We calculated estimated  glomerular filtration rates (eGFR) using the 4 variable modification of diet and renal disease (MDRD) equation and defined cases as those with an eGFR<60mL/min/1.73m2.  Presence of comorbid conditions was base on ICD-9 codes reported on the claims.  Propensity score methods were used to match controls to cases.  Data were analyzed using descriptive statistics and generalized linear models with a gamma distribution and log-link controlling for disease stage, health plan type, comorbid conditions, and demographic characteristics.  RESULTS: A total of 2,436 cases were identified with an eGFR of less than 60mL/min/1.73m2.   We were able to match controls to 2,223 (91.3%) of the identified CKD cases using propensity scores.  Unadjusted expenditures were $10,353 (sd 23,262) for CKD cases compared to $9,310 (sd 22,703) for matched controls.   After adjusting for health plan type, patient demographic characteristics, and comorbid conditions, adjusted expenditures were significantly greater for patients with CKD compared to matched controls depending on CKD stage.  Expenditures by stage ranged from 8% higher for patients with stage 3 disease to 62% higher for patients with stage 5 CKD compared to matched controls.  The key predictors of total treatment payments were disease stage, presence of comorbid conditions, age, and gender.  CONCLUSIONS: Payments for medical care attributable to CKD are significantly greater than medical care payments for those who do not have CKD and vary importantly by disease stage.  Interventions that prevent or delay the progression of CKD have the potential to save considerable resources.