OBJECTIVES: To validate the recently updated Charlson Comorbidity Index (CCI) for the prediction of future healthcare utilization for diabetic patients. METHODS: Administrative claims data were obtained for diabetic patients enrolled continuously for three years in the Department of Defense TRICARE program for retrospective analysis. The updated and the original CCI scores were calculated using baseline year data. Linear regression models were used to estimate log-transformed healthcare expenditures (COST) for one- and two-year post-index periods. Zero-inflated negative binomial regression models were used to estimate the number of hospitalizations (HOS) and the number of emergency department visits (ED) for one- and two-year post-index periods. The outcome variables were then dichotomized (above or below the 90th percentile of COST; ≥ 1 HOS or none; ≥ 1 ED or none) and estimated using logistic regression models. Adjusted R2, Akaike information criteria (AIC), and c statistics were assessed to compare the two CCI versions. RESULTS: A total of 8,704 patients were included in the study. The study population had a mean age of 51.0 years (SD: 10.5), and 46.3 percent were male. In the linear regression models, the updated CCI explained more variance than the original CCI in one-year COST (adjusted R2 = 13.0% vs. 11.3%) and two-year COST (adjusted R2 = 15.7% vs. 13.9%), adjusting for age and sex. The updated CCI was a better predictor of one- and two-year HOS (AIC = 8581, 13821) as well as one- and two-year ED (AIC = 17009, 25200). In the logistic regression models, the updated CCI performed better in predicting all study outcomes (c = 0.619 to 0.754) than the original CCI (c = 0.611 to 0.737). CONCLUSIONS: In a population of diabetic patients, the updated CCI showed improved predictive performance compared to the original CCI. The updated CCI should be validated in other patient populations.