OBJECTIVES: Frequently, commercial health insurance administrative claims databases are considered non-representative of the United States (US) population because they reflect only working age individuals, and their dependents who are currently employed. If this employment characteristic exists, it should be visible when large commercial administrative databases are compared against US Census demographic data. METHODS: This study compared the HealthCore Integrated Research Database (HIRD) and the MarketScan Database against US Demographic data for geographic region, age and gender. Age groups and geographic regions were coded to be consistent with US census divisions: Mid-West, West, North-East and South. Data were abstracted from 2009; the latest year that US Census data, HIRD and MarketScan data were available.  Patients were required to have generated at least 1 healthcare claim to be included in the analysis. RESULTS: The HIRD (n=14,794,609) and MarketScan (n=42,632,943) databases showed remarkable agreement between gender and age groups for all four geographic regions. Both MarketScan and the HIRD databases over represented US Census data for the age groups below 5 years and above 30 years of age. These agreement levels were consistent across all four regions and two gender groups. There was close agreement from ages 5 to 30 years for all sources. HIRD and MarketScan data under-represented the US Census data for age groups above age 65 years.     CONCLUSIONS: The employment characteristic is present in both the HIRD and MarketScan databases, and can be quantified against the US Census.  HIRD and MarketScan data are equally affected by the commercial database employment characteristic. The HIRD and MarketScan data agreement indicate that for research on specific disease or drug classification, it would be possible to extrapolate weighting data against US Census data to yield accurate estimates of disease prevalence among patients, or claims for coexisting and comorbid conditions and pharmacy claims.