OBJECTIVES: To investigate the possible influence of age on the refill-based adherence rates of anti-epileptic drugs METHODS: A retrospective drug utilisation review was performed on medicine claims data of a pharmacy benefit management company in South Africa. Refill-based adherence rates were calculated for 64,457 anti-epileptic drugs that were prescribed more than once during a four-year period (January 1, 2005 to December 2008). The refill-based adherence rate was calculated per trade name by using the following equation: Refill-Adherence rate = (total number of days of anti-epileptic drugs supplied – days supplied at the last refill) / (date last claimed – date first claimed). [RSA Rand (R)/$US = 6.38112 (2005); 6.78812 (2006); 7.06926 (2007) and 8.27505 (2008)].  RESULTS: Only 30.5% (n = 19 635) of anti-epileptic drugs had refill-adherence rates between between 90% and 110%. The majority of anti-epileptic drugs (58.9%; n = 37 962) had refill-adherence rates below 90% that accounted for 39.2% (n = R 57 599 838) of the total cost of all anti-epileptic drugs (N = R146 863 755) included in these calculations. Anti-epileptic drugs with refill-adherence rates > 110% (10,7%; n = 6 860) accounted for 6.7% (R9 782 864) of the total cost of all anti-epileptic drugs. The average refill-adherence rate decreased with nearly 10% from 93.7% in age group 1 (0 ≤ 12 years) to 83.3% in age group 5 (older than 65 years). In general the lowest average refill-based adherence rates were obtained with medicine items containing phenobarbitone and vitamin B1 (52.0% ± 37.8); phenobarbitone (63.5% ± 47.2); primidone metabolites (69.7% ± 47.2%); clonazepam (77.8% ± 184.8) and carbamazepine (80.9% ± 151.1)  CONCLUSIONS: Most of the anti-epileptic drugs had unacceptable low refill-adherence rates. Refill-adherence rates of anti-epileptic drugs decreased with an increase in the age of patients.