Approximately 30% of epilepsy patients experience seizures that are not controlled by the available drugs. Moreover, these drugs provide mainly a symptomatic treatment since they do not interfere with the disease's mechanisms. A mechanistic approach to the discovery of key pathogenic brain modifications causing seizure onset, recurrence and progression is instrumental for designing novel and rationale therapeutic interventions that could modify the disease course or prevent its development. In this regard, increasing evidence shows that neuroinflammation is a pathogenic factor in drug-resistant epilepsies. The High Mobility Group Box 1 (HMGB1)/Toll-like receptor 4 axis is a key initiator of neuroinflammation following brain injuries leading to epilepsy, and its activation contributes to seizure mechanisms in animal models. Recent findings have shown dynamic changes in HMGB1 and its isoforms in the brain and blood of animals exposed to acute brain injuries and undergoing epileptogenesis, and in surgically resected epileptic foci in humans. HMGB1 isoforms reflect different pathophysiological processes, and the disulfide isoform, which is generated in the brain during oxidative stress, is implicated in seizures, cell loss and cognitive dysfunctions. Interfering with disulfide HMGB1-activated cell signaling mediates significant therapeutic effects in epilepsy models. Moreover, both clinical and experimental data suggest that HMGB1 isoforms may serve as mechanistic biomarkers for epileptogenesis and drug-resistant epilepsy. These novel findings suggest that the HMGB1 system could be targeted to prevent seizure generation and may provide clinically useful prognostic biomarkers which may also predict the patient's response to therapy.