AIM: To assess the association between changes in medication and epilepsy-related hospitalisation. METHODS: Data were obtained from the PHARMO Record Linkage System (Jan 1998 to Dec 2002). We conducted a case-crossover study among patients with a frst epilepsy-related hospital admission who had continuously used at least one antiepileptic drug (AED) during a 28-week period before admission. For each patient, changes in medication in a 28-day window before hospitalisation were compared with changesin four earlier 28-day windows. Evaluated changes were: changes in AEDs (pattern and dosage), changes in interacting co-medication and changes in non-interacting co-medication (i.e. introduction of non-interacting drugs). The strength of the associationbetween changes in medication and epilepsy-related hospitalisation was estimated using conditional logistic regression analysis and expressed as odds ratios (ORs) with 95% confidence intervals (CI). RESULTS: Out of 1185 patients with a first epilepsy-related hospitalisation, 217 patients met the inclusion criteria. Of the changes in antiepileptic therapy, discontinuation showed a trend towards an increased risk of hospitalisation (OR: 2.57; 95%CI: 0.81-8.17). Drug interactions influencing antiepileptictherapy rarely occurred. Introduction of three or more non-interacting drugs was significantly associated with epilepsy-related hospitalisation (OR: 4.80; 95%CI: 2.12-10.87). Of individual drugs, addition of antimicrobial agents was significantly associated with epilepsy-related hospitalisation (OR: 1.99; 95%CI: 1.06-3.75). CONCLUSIONS: Changes in AED therapy were not significantly associated with epilepsy-related hospitalisation and few drug interactions influencing antiepileptic therapy occurred. However, patients starting three or more new non-AEDs had a nearly five times increased risk of epilepsy-related hospital admission. Copyright (c) 2006 John Wiley & Sons, Ltd.