Epilepsy affects 50 million persons worldwide, a third of whom continue to experience debilitating seizures despite optimum anti-epileptic drug (AED) treatment. Twelve-month remission from seizures is less likely in female patients, individuals aged 11-36 years and those with neurological insults and shorter time between first seizure and starting treatment. It has been found that the presence of multiple seizures prior to diagnosis is a risk factor for pharmacoresistance and is correlated with epilepsy type as well as intrinsic severity. The key role of neuroinflammation in the pathophysiology of resistant epilepsy is becoming clear. Our work in this area suggests that high-mobility group box 1 isoforms may be candidate biomarkers for treatment stratification and novel drug targets in epilepsy. Furthermore, transporter polymorphisms contributing to the intrinsic severity of epilepsy are providing robust neurobiological evidence on an emerging theory of drug resistance, which may also provide new insights into disease stratification. Some of the rare genetic epilepsies enable treatment stratification through testing for the causal mutation, for example SCN1A mutations in patients with Dravet's syndrome. Up to 50% of patients develop adverse reactions to AEDs which in turn affects tolerability and compliance. Immune-mediated hypersensitivity reactions to AED therapy, such as toxic epidermal necrolysis, are the most serious adverse reactions and have been associated with polymorphisms in the human leucocyte antigen (HLA) complex. Pharmacogenetic screening for HLA-B*15:02 in Asian populations can prevent carbamazepine-induced Stevens-Johnson syndrome. We have identified HLA-A*31:01 as a potential risk marker for all phenotypes of carbamazepine-induced hypersensitivity with applicability in European and other populations. In this review, we explore the currently available key stratification approaches to address the therapeutic challenges in epilepsy.