The 3 major seizure groups are focal onset, generalized onset, and unknown onset, classified by where seizure activity begins in the brain. Focal seizures start in one area of one hemisphere; generalized seizures engage both hemispheres from the start. Unknown onset covers seizures whose origin was never captured or confirmed. This three-group framework, set by the International League Against Epilepsy (ILAE) in 2017, replaced older labels like "partial" and "simple/complex" that confused patients and clinicians alike. It matters because onset type drives drug choice, workup, and driving restrictions.
What the 3 Major Seizure Groups Actually Classify
The 2017 ILAE classification sorts seizures by three sequential questions. First, where does the seizure begin: one hemisphere (focal) or both (generalized)? Second, is awareness impaired during the event? Third, does the seizure spread to involve both hemispheres secondarily?
Focal onset seizures are subdivided by awareness. A focal aware seizure preserves consciousness throughout, the patient is fully responsive. A focal impaired awareness seizure alters consciousness, and this is the group that produces the stereotyped automatisms most people picture: lip smacking, hand fidgeting, picking at clothing. Generalized onset seizures divide into motor and nonmotor types. Absence, myoclonic, tonic-clonic, tonic, atonic, and clonic seizures all sit here. Unknown onset is not a diagnosis of exclusion laziness; it is a deliberate category for events recorded on video without EEG confirmation, or for patients who cannot describe what happened.
In plain terms: the three groups answer "where did it start?" and everything else follows.
The practical separation between groups is not academic. Roughly 60 percent of seizures in adults are focal onset, while children show a higher proportion of generalized onset events, particularly absence and myoclonic types. That age split alone shapes the first drug a neurologist picks.
Focal vs Generalized Onset: The Defining Differences
The most consequential distinction is onset geography. Focal seizures arise in a network confined to one hemisphere, most commonly the temporal lobe (approximately 60 percent of focal cases) or frontal lobe. Generalized seizures begin in bilateral networks at once, without a detectable single focus on standard scalp EEG.
Three features separate them clinically:
- Duration: absence seizures typically last 5-10 seconds; focal impaired awareness seizures average 60-120 seconds; generalized tonic-clonic seizures run 1-3 minutes.
- Post-event state: generalized tonic-clonic seizures produce 15-30 minutes of postictal confusion and deep sleep; absence seizures show near-instant recovery within 5-10 seconds.
- Frequency: an individual with absence epilepsy can have 20-200 absence seizures per day; focal seizures in the same patient typically number 1-5 per week.
The comparison table below condenses the trade-offs across the groups.
| Feature | Focal Onset | Generalized Onset | Unknown Onset |
|---|---|---|---|
| Origin | One hemisphere, one network | Both hemispheres simultaneously | Unconfirmed or unwitnessed |
| Awareness | Preserved or impaired | Impaired (except some absence) | Often unknown |
| Typical duration | 60-120 seconds | 5-180 seconds by subtype | Variable |
| Postictal confusion | 5-30 minutes | 5-30 minutes (tonic-clonic); seconds (absence) | Variable |
| First-line drug class | Carbamazepine, levetiracetam, lamotrigine | Valproate, ethosuximide (absence), lamotrigine | Determined after EEG |
| EEG signature | Focal spikes or sharp waves | 3 Hz spike-and-wave (absence), generalized discharges | Normal or nondiagnostic in 30-50 percent |
Aura, Warning Signs, and the Pre-Seizure Timeline
Focal seizures frequently begin with an aura, a subjective warning that itself is a focal aware seizure. Auras are underreported: roughly 55-65 percent of temporal lobe epilepsy patients describe one, most commonly a rising stomach sensation, déjà vu, or a strange smell (uncinate fit). The aura phase lasts 5-30 seconds before awareness may fade.
The timeline of a typical focal-to-bilateral tonic-clonic event follows a predictable arc:
| Phase | Duration | What Happens |
|---|---|---|
| Prodrome | 30 minutes to 24 hours | Irritability, headache, sleep disruption |
| Aura | 5-30 seconds | Warning sensation, preserved awareness |
| Focal impaired awareness | 30-120 seconds | Automatisms, unresponsiveness |
| Secondary generalization | 30-90 seconds | Bilateral stiffening then rhythmic jerking |
| Postictal phase | 5-30 minutes | Confusion, fatigue, headache |
| Full recovery | 1-24 hours | Gradual return to baseline |
The prodrome is the least understood phase. Per a 2020 review in Epilepsy & Behavior, only about 30 percent of patients recognize prodromal symptoms for what they are, and fewer still can act on them. This is where rescue medications such as intranasal midazolam or buccal lorazepam can be deployed, with a 10-20 minute window before a seizure becomes prolonged.
How Seizures Spread: The Mechanism Behind Each Group
Seizures arise from an imbalance between excitation and inhibition, specifically excessive glutamate signaling coupled with reduced GABAergic inhibition. In focal epilepsy, a small cluster of hyperexcitable neurons fires repetitively at 3-8 Hz, recruiting neighbors until a seizure focus forms. This local event becomes visible on EEG before any clinical symptom, a phenomenon called electrographic onset.
Generalized seizures skip the focal recruitment step. In absence epilepsy, thalamocortical circuits oscillate at a fixed 3 Hz, driven by T-type calcium channels in thalamic relay neurons. This is why ethosuximide, which blocks T-type calcium channels, works for absence seizures but not for focal seizures. The channel specificity is not incidental, it reflects the different mechanism of each group.
Here is the non-obvious correction: a seizure does not "spread across the brain like a wave" in all cases. In generalized onset events, bilateral involvement is present at onset, not the result of propagation. Misreading a generalized seizure as focal spreading is a common error and can lead to the wrong drug class. An estimated 20-30 percent of misclassified cases result from this confusion, per a 2019 study in Epilepsia.
Red Flags: When a Seizure Group Demands Emergency Care
Most seizures end spontaneously. Certain patterns require emergency evaluation.
- Any seizure lasting more than 5 minutes, or multiple seizures without recovery between them, constitutes status epilepticus and carries a 15-30 percent mortality rate if untreated.
- A first-ever seizure in an adult over age 40 warrants urgent imaging; structural causes such as tumors or stroke account for 20-30 percent of new-onset cases in this age group.
- A focal seizure with persistent weakness after the event (Todd paresis) lasting over 30 minutes may indicate stroke and requires urgent assessment.
- A seizure occurring during pregnancy in a patient with preeclampsia points to eclampsia, which carries a 1-2 percent maternal mortality risk without treatment.
- Seizures with fever in an adult, or new focal features in a known epilepsy patient, suggest infection or lesion progression.
- A seizure followed by a prolonged postictal state exceeding 60 minutes may signal nonconvulsive status epilepticus.
In plain terms: five minutes is the line. Cross it and call emergency services.
What to Actually Do: Matching Action to Seizure Group
The right response depends on which of the 3 major seizure groups you are facing.
For focal aware seizures, stay with the person, speak calmly, and time the event. Most resolve within two minutes. Note the features for the neurologist: the specific aura, automatisms, and duration. These details narrow the focus and reduce the need for inpatient video-EEG monitoring, which costs USD 3,000 to 8,000 per admission.
For focal impaired awareness or generalized tonic-clonic seizures, protect the head, clear hard objects, turn the person on their side once jerking stops, and never place anything in the mouth. Time from onset. Rescue medication protocols, typically intranasal midazolam at 5 mg for adults, are appropriate after 5 minutes if prescribed.
For absence seizures, no physical intervention is needed. The event ends on its own within 10 seconds. Log frequency if possible; more than 50 per day suggests inadequate dosing.
For unknown onset seizures, the priority is capture. A smartphone video of the event is worth more than a week of outpatient testing. Roughly 40 percent of unknown onset seizures are reclassified after a single witnessed video.
Practical commitments that change outcomes:
- Keep a seizure diary with timestamps, duration, and triggers (missed sleep, alcohol, missed doses).
- Confirm drug adherence; missed doses account for 30-50 percent of breakthrough seizures.
- Ask for EEG within 72 hours of a first event; yield drops after that window.
- Review driving restrictions with the neurologist; most US states require 3-6 months seizure-free.
Limits of the Three-Group Model
The ILAE 2017 classification is the current standard, but it is a framework, not a complete neurobiological map.
First, roughly 25-30 percent of patients with epilepsy are drug-resistant, and their seizures often blur the group boundaries, with focal events that mimic generalized patterns on EEG. Second, scalp EEG misses deep focal sources in the temporal or frontal lobe in up to 50 percent of cases, so some "unknown onset" seizures are simply focal events the electrodes cannot see. Third, genetic generalized epilepsies occasionally present with focal features, contradicting the clean split the groups imply.
The classification also does not predict outcome. Two patients in the same group can have radically different prognoses depending on etiology, age at onset, and lesion status. The next revision of the model, anticipated after the 2025 ILAE consensus meetings, may add an etiology axis to each group. Until then, treat the 3 major seizure groups as a starting vocabulary, not a final diagnosis.
What the labels largely ignore is semiology, the actual behavior of the seizure. A neurologist watching a 90-second video can often localize a seizure better than a scalp EEG. That gap between classification and observation is the real edge case clinicians manage daily.