OBJECTIVES: VA (Veteran's Health Administration) Cooperative studies convincingly demonstrate that despite equal efficacy, phenobarbital and phenytoin are more likely to cause adverse effects, and an expert panel recommended carbamazepine as the drug of choice for adult-onset seizures, with newer medications (lamotrigine, gabapentin) being preferable for treating elderly epileptics. This study examines the state of AED use in the VA during 1999. METHODS: We linked administrative (FY1997-FY1999) and pharmacy databases to identify veterans with epilepsy who were on anti-epileptic drugs (AEDs, N=63,853). We used logistic regression to determine if age (>65 vs. < 65) or type of care (primary care or primary care and neurology) predicted use of: phenobarbital, phenytoin, carbamazepine, and gabapentin. We controlled for demographics, year diagnosed (1997 or before, 1998, 1999), seizure type (partial, generalized, both), and number of epilepsy hospitalizations or emergency visits (severity). RESULTS: In 1999 patients were on: phenytoin 59%, phenobarbital 10%, carbamazepine 20%, gabapentin 12%, lamotrigine 2%. Logistic regression indicated that when year diagnosed, sex, race, and severity were controlled, veterans under 65 were less likely to receive phenobarbital (OR: .52-.56) and phenytoin (OR: .46-.49), and more likely to receive carbamazepine (OR: 1.67-1.80 ) and gabapentin (OR: 1.56-1.69). Those seen by neurology were less likely to receive phenobarbital (OR: .68-.75) and phenytoin (OR: .73-.82), and more likely to receive gabapentin (OR: 1.58-1.88) and carbamazepine (OR: 1.28-1.35). CONCLUSIONS: The elderly and those seen only in primary care were more likely to receive potentially problematic medications. Despite research findings, physicians may be reluctant to change efficient medications. The elderly tend to have more adverse effects from these medications, and while often subtle, they may result in injury and debilitation. Results highlight the need translate clinical research into best practices, and develop research programs that relate these practices to patient outcomes.