Objective: Medication non-adherence among patients with chronic conditions can have both clinical and economic consequences. The objective is to investigate whether non-adherence to antiepileptic drugs (AEDs) is associated with increased morbidity relative to adherence, as proxied by healthcare utilization and costs, in a Medicaid population with epilepsy. Methods: A retrospective open-cohort design using state Medicaid claims data from Florida, Iowa and New Jersey in the period of January 1997-June 2006 was employed. Patients aged = 18 with = 1 diagnosis of epilepsy, = 1 neurologist visit, = 2 AED dispensings, and = 6 months of baseline period were included. Medication possession ratio (MPR) was used to evaluate AED adherence on a quarterly basis with MPR = 0.8 considered adherent and <0.8 non-adherent. The association of non-adherence with health care utilization was assessed using univariate and multivariate Poisson regressions to model frequency of hospitalizations, inpatient days, emergency room (ER), and outpatient visits per person-year of observation. Quarterly per-patient inpatient, outpatient, ER, and pharmacy costs were calculated across non-adherent and adherent quarters for the under-65 population and cost differences computed. Adjusted incremental costs of non-adherence were estimated with multivariate Tobit regression models. Results: A total of 33,658 patients met the study inclusion criteria (28,470 under-65), together contributing 388,564 (74%) adherent and 136,550 (26%) non-adherent quarters. Non-adherence was associated with significantly higher incidence of hospitalizations (incidence rate ratio [IRR] =1.39, 95% confidence interval [CI]=1.37-1.41), inpatient days (IRR=1.76, 95% CI=1.75-1.78), and ER visits (IRR=1.19. 95% CI=1.18-1.21). Non-adherence was associated with positive quarterly incremental costs related to serious outcomes, including inpatient ($4320, 95% CI=$4077-$4564) and ER ($303, 95% CI=$273-$334) services. Conclusion: Non-adherence to AEDs is relatively common and appears to be associated with increased morbidity as represented by higher health care utilization and costs.