Can Seizures Stop on Their Own? The Science of Self-Termination and When to Act

Yes, most seizures stop on their own. Roughly 7 in 10 single seizures end within 2 to 3 minutes without any medication, because the brain has built-in braking systems that actively shut down runaway electrical activity. The critical caveat is the 5-minute mark: per the 2015 International League Against Epilepsy definition, a convulsive seizure still running at 5 minutes is status epilepticus and needs emergency treatment, because the same braking systems that usually work begin to fail the longer a seizure lasts.

The Braking System: How the Brain Ends a Seizure

A seizure is a self-sustaining loop of synchronous firing. For it to stop, the brain must actively break that loop, and it does so through several layers of defense operating on different timescales.

The first layer is cellular exhaustion. Neurons that fire rapidly progressively lose their ability to keep firing: sodium channels enter an inactivated state, potassium conductance rises, and the cell's own afterhyperpolarization pushes it below the threshold for firing. This is not a passive rundown; it is an active, use-dependent brake that engages within seconds to a minute of intense firing.

The second layer is chemical. Each seizure discharge releases inhibitory neurotransmitters, chiefly GABA, onto the very circuits that are seizing. Simultaneously, the energy molecule adenosine accumulates in the extracellular space and binds A1 receptors, which suppress excitatory transmission. Adenosine builds up faster during high-frequency firing, so the brake strengthens precisely when the seizure is most intense, a negative-feedback loop exquisitely matched to seizure duration.

The third layer is network-level. Inhibitory interneurons, a minority of cells that nonetheless control the timing of entire populations, are recruited during the seizure and begin firing in patterns that fragment the synchronous loop. Subcortical structures, including the substantia nigra pars reticulata and thalamic nuclei, also feed suppression back into cortex. Animal work shows that disrupting these circuits prolongs seizures, which is why the braking system is best described as an active control network rather than a simple fatigue process.

In plain terms: the brain does not wait for a seizure to burn out. It deploys cellular, chemical, and network brakes simultaneously, and in most seizures one or more of them wins within a couple of minutes. Why the system fails in the remaining cases is the subject of the next section.

Why Some Seizures Can't Stop: The 5-Minute Threshold

The brakes sometimes lose. When seizure activity persists, two things happen that make termination progressively harder.

First, GABA receptors retreat. Within minutes of sustained seizure activity, neurons internalize their GABA-A receptors, pulling them off the cell surface and into the cytoplasm. Fewer receptors means weaker inhibitory signaling exactly when the brain needs it most. This is why benzodiazepines, which work by enhancing GABA-A receptor function, become less effective the longer a seizure runs; the drug's target is literally disappearing from the membrane. It is also the mechanistic reason emergency plans emphasize early treatment rather than watchful waiting.

Second, the seizure becomes self-reinforcing. Repeated firing strengthens excitatory synapses and recruits more tissue into the discharge, so the network sustaining the seizure grows even as the inhibitory response grows. The race between braking and spread is roughly even for the first few minutes, and after that the seizure side increasingly wins.

The numbers around this threshold are stark. Status epilepticus, defined operationally as a convulsive seizure lasting 5 minutes or more, affects an estimated 12 to 20 per 100,000 people per year in the United States, or roughly 40,000 to 60,000 episodes annually, though European cohorts report higher rates approaching 25 per 100,000 when all seizure types are counted. Mortality from a first episode of convulsive status epilepticus ranges from about 7 to 15 percent in population-based studies, and rises sharply with age and with the duration of seizures before treatment. Roughly 4 in 10 people who present in status epilepticus have no prior epilepsy diagnosis; the seizure is the first sign of a new problem, which is why a first prolonged seizure demands urgent evaluation rather than reassurance.

One clarification prevents a common contradiction: you will see status epilepticus defined as 5 minutes in clinical guidelines and as 30 minutes in older research literature. Both are correct for their purpose. The 5-minute operational definition exists because self-termination becomes unlikely and treatment should begin; the 30-minute definition describes the point at which lasting neuronal injury is established. The modern standard, per the 2015 ILAE position, is to treat at 5 minutes and to treat aggressively before 30.

How Long Each Seizure Type Lasts Without Treatment

Self-termination is the rule across seizure types, but the timelines differ enough to matter for decisions. The table below summarizes typical untreated durations and the probability that a single episode stops without intervention.

Two entries deserve a closer look, because they are where misconceptions cluster. Absence seizures are almost always self-limiting, which leads families to dismiss them; but a child having 50 brief absence seizures a day is losing minutes of awareness hourly, and the untreated pattern can affect learning and safety even though no single seizure is dangerous. Conversely, generalized tonic-clonic seizures are the ones that most often stop on their own, yet they are also the ones that most often fail to stop, and they carry the highest risk of injury and of progression to status epilepticus. Self-termination rates and danger do not track together, which is why duration and seizure type must be assessed together.

Where Seizures Stop on Their Own but Care Still Can't Wait

The most useful distinction in seizure care is not "stops on its own" versus "doesn't." It is "stops on its own and needs no emergency care" versus "stops on its own but still needs urgent evaluation." A seizure can end unassisted and still be a medical red flag.

First seizures are the clearest example. A first-ever convulsion that stops after 90 seconds may signal a structural lesion, a metabolic derangement, an infection, or an epilepsy syndrome, and guidelines from the American College of Emergency Physicians and the American Epilepsy Society support urgent evaluation after any first seizure, ideally within hours to days depending on the clinical picture. The seizure stopping is not evidence that the cause is benign.

Seizure clusters are a second example. Defined as two or more seizures within 24 hours, clusters occur in roughly 30 to 40 percent of people with epilepsy at some point, and they predict progression to status epilepticus. Each individual seizure in a cluster may stop on its own, but the pattern itself is the warning, and rescue medication protocols exist precisely for this scenario.

Prolonged postictal states are a third. After a seizure ends, the brain is exhausted: confusion, drowsiness, headache, and muscle soreness commonly last 10 to 30 minutes, and can last hours after a convulsion. A postictal state that deepens instead of lifting, or that includes difficulty breathing, persistent unconsciousness beyond a few minutes, or a new focal deficit such as weakness on one side, warrants emergency assessment even though the seizure itself is over.

Finally, psychogenic non-epileptic seizures deserve mention because they are routinely misjudged by duration alone. These episodes can last 5 to 20 minutes or longer, far beyond typical epileptic seizures, which leads bystanders to assume a dangerous convulsion is occurring. They are not treated with anti-seizure medication and require diagnostic video-EEG for clarity, and the person is not at risk of the same neuronal injury as in status epilepticus, though the underlying condition needs care. Duration alone cannot distinguish the two, which is why witnessed description and medical evaluation matter.

The Conditions Under Which Self-Termination Actually Fails

Most seizures stop on their own, but several conditions reliably shift the odds against self-termination, and knowing them helps families and clinicians anticipate rather than react.

Sleep deprivation is one of the best documented. It lowers seizure thresholds and, in people with epilepsy, increases both seizure frequency and the likelihood of clusters. Missed medication is another: sudden withdrawal of anti-seizure drugs, particularly benzodiazepines and barbiturates, can trigger seizures that are unusually prolonged and resistant to termination.

Alcohol withdrawal follows the same logic, producing seizures that often cluster and can progress to status epilepticus. CNS infections, acute stroke, traumatic brain injury, and metabolic disturbances such as severe hypoglycemia or hyponatremia create a brain state in which the braking system is compromised, and seizures in these settings stop on their own less often and recur more.

The most important stratum is treatment history. People with well-controlled epilepsy whose seizures have always been brief have a high probability of continued self-termination. People with a history of status epilepticus, prior prolonged seizures, or refractory epilepsy do not, and their emergency plans should assume the brakes may fail. This is not fatalism; it is calibration, and it is the reason personalized seizure action plans exist.

One counterintuitive detail is worth stating plainly: a seizure that stops on its own does not become safer because it was brief. Brief seizures still carry risks of falls, burns, and traffic injury if they occur in unsafe settings, and repeated brief seizures are associated with cumulative cognitive and psychosocial burden. The reason to treat epilepsy is not only to prevent status epilepticus; it is to prevent the smaller harms that ordinary seizures inflict month after month.

What to Actually Do When a Seizure Starts

The single most useful action is to time the seizure and protect the person while waiting for the expected self-termination. The following list covers the concrete steps, ordered by urgency.

• Time it. Note the clock time when the seizure begins. The 5-minute threshold is the decision point for emergency medication and emergency services, and an accurate time is the most valuable information you can give responders.
• Protect the head and clear the space. Move furniture and sharp objects away, place something soft under the head, and do not restrain the person. Restraint increases injury risk and does not shorten the seizure.
• Turn the person onto their side once the convulsion ends. Recovery position clears the airway and prevents aspiration; this matters more than most people realize and is the main reason postictal breathing sounds are usually harmless.
• Do not put anything in the mouth. The tongue cannot be swallowed, and objects cause dental injury and airway obstruction.
• Call emergency services if the seizure lasts 5 minutes or more, if a second seizure follows within 24 hours, if the person is injured, pregnant, in water, or has no known epilepsy history, or if breathing does not normalize after the seizure. When in doubt, call.
• Use rescue medication if prescribed and trained. Rectal diazepam or intranasal midazolam, per a clinician's individualized plan, is designed for seizures lasting longer than 3 to 5 minutes, and earlier administration is more effective because GABA receptors are still on the cell surface.

After the seizure, the practical timeline continues. Stay with the person until full awareness returns, which typically takes 10 to 30 minutes after a convulsion and longer after a cluster. Document what happened, including duration, what the person was doing at onset, whether they lost awareness, and how they behaved afterward. That description is often more diagnostic than any test, and it directly informs whether the next step is routine follow-up or urgent assessment.

Medication adherence, sleep, and alcohol moderation remain the three highest-yield levers for reducing the chance that a future seizure fails to stop on its own. They are unglamorous, but they change the odds more than any bystander action can.

Limits of the Evidence and What Remains Uncertain

The claim that most seizures stop on their own is well supported, but the precise probability is less certain than headlines suggest. Estimates of untreated termination rates come largely from video-EEG monitoring units and emergency cohorts, which oversample severe cases; community-based data are sparser. The commonly cited figure that about 70 percent of single convulsive seizures end within 2 to 3 minutes is a reasonable synthesis, not a universal law.

The mechanisms of termination are also partly inferred. Adenosine's role is established in animal models and supported by human cerebrospinal fluid studies, but the relative contribution of each braking system in any individual seizure is not measurable at the bedside. The GABA receptor internalization story is solid pharmacology, yet the exact kinetics in human status epilepticus are extrapolated from limited tissue data.

What is settled, and what should govern decisions, is the clinical frame: brief seizures usually end unassisted; the 5-minute threshold marks the point where self-termination probability falls and treatment should begin; and a first seizure, a cluster, or a seizure with a new cause deserves urgent evaluation even when it stops. Those three rules are stable across guidelines and are enough to act on. The open questions concern how to predict, in advance, which individual seizure will be the one that doesn't stop, and that remains an active area of research rather than a settled answer.