Objective:  To develop a methodology to link patients from two de-identified databases and leverage unique data from both to measure the impact of blood pressure and clinical findings on total costs.  Methods:  Hypertensive patients (ICD-9 diagnosis 401.xx-405.xx) were identified from the MarketScan Commercial and Medicare Supplemental administrative claims databases (MarketScan) and the GE Centricity Electronic Medical Record (EMR) database (Centricity) for the years 2004-2008.  A hybrid approach  of deterministic and probabilistic matches was developed to identify common patients.  Patients were included if they matched on zip code, gender and month of birth, and had at least three matching office visit dates at a rate of 75% or higher.  Patients were followed for 12 months after the initial diagnosis.  MarketScan provided data on enrollment, all reimbursed services (medical and drug) and costs, and Centricity provided clinical and biometric details, such as body mass index (BMI) and blood pressure.  Results:  Among the 3 million MarketScan and 1.5 million Centricity patients with hypertension, 31,786 met the matching criteria.  Mean age was 58 and 54% were female.  The demographic and clinical characteristics of these patients did not vary substantially from those of the two data sources.  Among the 31,786 patients, 84% received drug treatment, 56% had a BMI over 30 and mean systolic and diastolic values were 134 and 81, respectively.  Mean unadjusted costs were $9,338 for patients with consistently controlled (first and last systolic <140 and diastolic<90) hypertension and $8,773 for patients not consistently controlled.  Conclusions:  A combined probabilistic and deterministic approach of linking patients yielded a sample size large enough to conduct a study and leverage the strengths of administrative and EMR data.  Initial findings suggest that controlled patients incur higher costs, however, adjustments have not been made for additional demographic, clinical, and treatment characteristics.