OBJECTIVE: The nation's largest single provider of healthcare to those infected with Human Immunodeficiency Virus (HIV) is the Department of Veterans Affairs (VA). The clinical effectiveness of highly active antiretroviral therapy (HAART) depends considerably on an individual's ability to adhere closely to therapy. The objective of this study was to assess the relationship between patient adherence to HAART and the cost-effectiveness of HAART in antiretroviral-naïve HIV-infected VA patients. METHODS: The overall design of the study is a hybrid between an observational study and a decision-analytic model. Kaplan-Meier survival estimates, calculated from the VA's laboratory and clinical data, were used to model HIV-disease progression. Pharmacy-based adherence was calculated using refill percent adherence obtained from prescription records; individuals were categorized as adherent if adherence was equal to or greater than 95%. Inpatient and outpatient costs were derived from the VA Health Economics Resource Center estimates. Outpatient prescription costs were obtained from the VA's Pharmacy Benefits Management database. Markov Monte Carlo simulations were conducted to project the cost-effectiveness of adherence to HAART. Costs and quality-adjusted life-years (QALY) were discounted at 3%. RESULTS: For the five-year Markov Monte Carlo simulations, the overall costs of adherent behavior were lower than those of non-adherent behavior. In addition, the projected QALYs of adherent behavior were greater than those of non-adherent behavior; therefore, the cost-effectiveness of non-adherent behavior was dominated by adherent behavior. For the 20-year simulations, the cost-effectiveness of adherent behavior, compared to non-adherent behavior, was less than $15,000/QALY. CONCLUSIONS: Medication non-adherence in chronic conditions is a recognized public health problem. The association between medication adherence and improved health outcomes is well documented and the results of this study provide further evidence of this relationship. In this study, the marginal costs and effects associated with adherent behavior, compared to non-adherent behavior, are well below acceptable willingness-to-pay thresholds.