OBJECTIVES: The validity of diagnosis/procedure coding for determining the severity of renal impairment is unknown. This retrospective, observational study developed an administrative claims-based algorithm which classified patients with type 2 diabetes mellitus (T2DM) into renal impairment stages using estimated glomerular filtration rate/1.73 sq M by MDRD equation (eGFR) as the measure for renal function. METHODS: The data source was U.S. administrative claims collected from among a sample of 35,624 patients ≥18 years of age who, during the period from 1/1/2012-12/31/2012, had ≥1 laboratory result for eGFR, ≥2 medical claims with a diagnosis code for T2DM, continuous insurance enrollment, and no medical claims with a diagnosis/procedure code for type 1 diabetes, gestational diabetes, or pregnancy. The sample was divided into two equal random samples: a test set and validation set. Among the test set, four logistic regressions were fit modeling Kidney Disease Outcomes Quality Initiative-defined renal impairment stages (eGFR <15, <30, <60, and <90) as a function of age, sex, and 25 binary indicators for the presence of medical claims with renal impairment-related diagnosis/procedure codes. From each regression, a predicted probability was obtained for the validation set and performance of the algorithm was tested (e.g., by ROC analysis) at varying probability cutoff classification thresholds. RESULTS: In the validation set, the percentage of patients correctly classified by the test set algorithm using a standard probability cutoff=0.5 was 75.9% for eGFR <90, 82.1% for <60, 97.3% <30, and 99.3% for <15; in the test set, these same percentages deviated by less than 1 percentage point. Model C-statistics ranged from 0.79 for eGFR <90 to 0.89 for eGFR <15. Sensitivity/specificity varied considerably by selected probability cutoffs. CONCLUSIONS: This novel, replicable, administrative claims-based algorithm should prove useful to diabetes researchers who need to classify patients’ renal impairment stage in the absence of detailed eGFR data.