OBJECTIVES: About one-third of patients with epilepsy may be refractory to antiepileptic drug (AED) treatment. This study compared healthcare utilization and costs (all-cause and epilepsy-related) associated with refractory and non-refractory epilepsy. METHODS: Texas Medicaid prescription and medical claims from 09/01/07-12/31/13 were analyzed over a 30-month study period. Patients eligible for the study: 1) were 18-62 years of age at index, 2) had an AED claim during the identification period (03/01/08-12/31/11) with no 6-month baseline AED use, and 3) had evidence of epilepsy diagnosis during the study period. Based on AED use in the identification period, patients were categorized as “refractory” (≥3AEDs) and “non-refractory” (<3AEDs). The index date was the date of the first AED claim. Healthcare utilization and costs included medical service claims comprising of inpatient hospitalizations, emergency department (ED) and outpatient visits, and pharmacy claims. Patients in both cohorts were matched 1:1 using propensity scoring and further compared for healthcare utilization and costs using non-parametric methods. RESULTS: Of the 10,599 eligible patients, 2,798(26.4%) patients in the refractory cohort were matched 1:1 to patients in the non-refractory cohort. Mean unadjusted all-cause and epilepsy-related annual costs (2013 USD) to the Texas Medicaid program were approximately 17% and 101% greater, respectively, for refractory vs. non-refractory patients (all-cause: $23,136 vs. $19,813; epilepsy-related: $7,811 vs. $3,893). After controlling for covariates, patients with refractory epilepsy had a significantly greater number of epilepsy-related hospitalizations, ED and outpatient visits, and pharmacy claims, as well as higher costs in each category, than their non-refractory counterparts (p<0.0001).  Utilization and costs were also significantly greater for all-cause outpatient visits (p<0.05) and pharmacy claims (p<0.0001) for the refractory cohort. CONCLUSIONS: Results highlight the importance of better management of epilepsy, especially refractory epilepsy, which may significantly decrease the utilization of healthcare resources and, thereby, decrease costs to Medicaid programs.