Not all seizures show up on an EEG, and the gap between what is happening in the brain and what a standard scalp recording captures is wider than most people realize. A routine EEG performed on an adult after a first unprovoked seizure picks up abnormal electrical activity only about 17% of the time, according to pooled data across studies. The reasons range from simple timing to the physics of how electrical signals travel through skull and scalp tissue, and understanding those reasons matters if you or someone you know has had a seizure but been told the EEG looked “normal.”
How Often a Routine EEG Catches Something
A routine EEG typically runs for about 20 to 30 minutes. For adults who have had a single unprovoked seizure, the pooled sensitivity of that short recording is around 17%, with high specificity near 95%. In children, the numbers look better but are still far from perfect: sensitivity is roughly 58%, with specificity around 70%.1PubMed. The diagnostic accuracy of routine electroencephalography after a first unprovoked seizure That means for an adult, a normal routine EEG after a seizure is the rule, not the exception. The test’s strength lies more in what an abnormal result tells you than what a normal result rules out.
This low sensitivity is not because the technology is broken. It reflects several overlapping problems: seizures are brief events, EEG only records what is happening at the moment the electrodes are on, and the electrical signals that indicate epilepsy between seizures (called interictal spikes) do not fire on command. Think of it as trying to photograph lightning with a 20-minute time-lapse in a storm that might last all week. You might catch it, or you might just get pictures of clouds.
Why the Skull Gets in the Way
Scalp EEG does not read the brain directly. Electrodes sit on top of skin, muscle, and bone, all of which blur and weaken the electrical signals radiating outward from brain tissue. For a spike of abnormal electrical activity to be visible on a scalp recording, it needs to involve a fairly large patch of cortex firing in sync. Research using simultaneously placed intracranial and scalp electrodes has shown that when a cortical spike covers more than about 10 square centimeters of brain surface, it shows up on scalp EEG roughly 90% of the time. But when the source area is smaller than 10 square centimeters, only about 10% of those spikes make it through to the scalp. Spikes originating from patches smaller than about 6 square centimeters were never visible on scalp EEG at all.2PubMed. Intracranial EEG substrates of scalp EEG interictal spikes
This matters because many seizures begin in a small focal area before spreading. If the seizure origin is compact and stays that way, or if the interictal discharges between seizures are generated by a small cluster of neurons, the scalp EEG may show nothing at all. The signal simply does not reach the surface with enough strength to stand out from background brain activity.
Deep Brain Structures Are Especially Hard to Record
Location matters as much as size. The brain is not a smooth ball; it has deep folds and buried structures. One of the trickiest areas is the insula, a region of cortex tucked deep within the lateral sulcus, essentially hidden behind the temporal and frontal lobes. Seizures that originate in the insula often produce no clear signal on scalp EEG, or they produce signals that look like they are coming from somewhere else entirely, such as the temporal or frontal lobe.3PubMed Central. Insular epilepsy, an under-recognized seizure semiology. A review for general neurologist This can lead to mislocalization, where the EEG points doctors toward the wrong part of the brain.
Interictal and ictal EEG patterns from the insula are described as “variable and misleading,” which is a polite way of saying the standard test can actively point in the wrong direction for these patients.4PubMed Central. The Insula and Its Epilepsies Insular epilepsy is considered under-recognized partly because the tool most commonly used to investigate seizures is not well suited to detecting it. Other deep or medial brain regions can pose similar challenges, though the insula is the most studied example.
Seizures arising from these deeper structures are sometimes called “scalp-negative seizures,” meaning they are genuine electrographic seizures that produce no specific correlate on a standard scalp recording.5Scientific Archives. Negative is Not Always Negative: Improving Outcomes in Scalp Negative Seizures Using Intracranial EEG The seizure is real, the brain is seizing, but the surface test cannot see it.
Muscle Artifact During Convulsive Seizures
Here is an irony that surprises many people: the more dramatic the seizure, the harder it can be for EEG to read the brain’s electrical activity during it. In a generalized tonic-clonic seizure, the kind with full-body stiffening and shaking, the intense muscle contractions generate their own electrical signals. Those muscle signals are picked up by the same scalp electrodes and can completely drown out the brain waves underneath. One study found that muscle artifacts contaminated 97% of ictal EEGs (recordings made during an actual seizure) and interfered with interpretation in 76% of cases. The problem was worse with seizures originating outside the temporal lobe.6PubMed. Improving the interpretation of ictal scalp EEG: BSS-CCA algorithm for muscle artifact removal
Generalized tonic-clonic seizures are especially difficult to analyze on scalp EEG precisely because of this muscle artifact problem.7Electroencephalography and Clinical Neurophysiology. Searching for hidden information with Gabor Transform in generalized tonic-clonic seizures Specialized signal-processing techniques can help strip away the muscle noise after the fact, but in real-time clinical monitoring, the convulsive phase of a seizure often looks like a mess of artifact on the screen. Doctors still see enough before and after the convulsion to confirm what happened, but the moment-to-moment details during the shaking phase may be unreadable.
How Longer Monitoring Changes the Picture
Given that a standard 20-to-30-minute EEG misses most abnormalities in adults, one natural question is whether recording for longer helps. The answer is clearly yes. Ambulatory EEG, where you wear a portable recording device for a day or more as you go about your normal life, significantly improves the chance of capturing both epileptic and non-epileptic events compared to a routine in-office recording.8PubMed. Clinical findings of long-term ambulatory video EEG following routine EEG
In hospital-based continuous monitoring, a retrospective study found that about 69% of subjects showed epileptic abnormalities during video-EEG monitoring, with the first abnormality appearing on average around 8 hours in. In the vast majority of those cases, the abnormality was detectable within the first 24 hours. But in about 5% of all subjects, epileptic activity did not appear until after 24 hours of continuous recording. And among patients who had actual seizures during monitoring, nearly a third of those seizures occurred after the 24-hour mark.9PubMed Central. Assessing 72 h vs. 24 h of long-term video-EEG monitoring to confirm the diagnosis of epilepsy: a retrospective observational study This is one reason epilepsy monitoring units often keep patients hooked up for several days.
Another study looked at what happens when a routine EEG comes back nondiagnostic and the patient is then moved to extended outpatient monitoring. The routine EEG alone was diagnostic in about 27% of patients. Among those whose routine EEG was nondiagnostic, extended monitoring picked up a diagnosis in an additional 32%.10PubMed. Diagnostic yield of sequential routine EEG and extended outpatient video-EEG monitoring The extended sessions were particularly useful for catching focal seizures and psychogenic nonepileptic events, both of which tend to slip through short recordings.
Sleep Deprivation and Repeat Testing
If a first EEG is normal, doctors sometimes order a second one performed after partial sleep deprivation. The idea is that the transition from wakefulness to sleep can provoke interictal discharges that would not appear during a routine awake recording. In a study of children with newly diagnosed seizures, a repeat EEG after partial sleep deprivation picked up epileptic abnormalities in about 35% of cases. For the overall group, this added 11 percentage points of detection on top of the 56% who already had abnormalities on the initial standard EEG. Roughly half of the abnormalities on the repeat test appeared only during sleep.11PubMed. The diagnostic yield of a second EEG after partial sleep deprivation: a prospective study in children with newly diagnosed seizures
Sleep deprivation EEGs are a common next step when the clinical suspicion for epilepsy is strong but the first recording was unrevealing. They are not a guarantee, but they meaningfully improve the odds. People with generalized epilepsy tend to show interictal abnormalities faster than those with focal epilepsy, so the type of epilepsy suspected also influences how likely any given EEG is to be informative.12Seizure. Latency to first spike in the EEG of epilepsy patients
Nonconvulsive Seizures in the ICU
One situation where the gap between clinical appearance and EEG findings runs in the opposite direction is the intensive care unit. Here, patients may be having seizures with no visible outward signs at all. These are called nonconvulsive seizures: the brain is seizing electrically, but the patient might just look sedated or confused, or show no change in behavior whatsoever. Nonconvulsive seizures occur in roughly 10 to 20% of ICU patients who undergo continuous EEG monitoring. The risk is highest in people who are comatose, in pediatric patients, and in those with structural brain injuries or a history of prior seizures.13PubMed. EEG in the critical care setting
In one large study of critically ill patients, seizures were detected in 19% of those monitored with continuous EEG. Of the patients found to be seizing, 92% had exclusively nonconvulsive seizures, meaning no one would have known about them without the EEG.14PubMed. Detection of electrographic seizures with continuous EEG monitoring in critically ill patients This flips the original question on its head: in the ICU, the EEG is often the only way to know a seizure is happening, because the body gives no external clue.
Even with continuous monitoring, interpretation is not always straightforward. Some EEG patterns in critically ill patients fall on a gray zone sometimes called the ictal-interictal continuum, patterns that look potentially harmful and may represent subtle ongoing seizure activity but do not meet the classic criteria for a seizure. There are no universally agreed-upon guidelines for how to manage these ambiguous patterns.15PubMed. Understanding and Managing the Ictal-Interictal Continuum in Neurocritical Care
Frontal Lobe Seizures and Psychogenic Events
Frontal lobe epilepsy presents a particularly frustrating diagnostic puzzle. Seizures originating in the frontal lobes often produce dramatic motor behaviors, such as thrashing, cycling leg movements, or sudden posturing, that can look bizarre and emotional. The problem is twofold: these seizures frequently produce no identifiable changes on scalp EEG, and their outward appearance closely resembles psychogenic nonepileptic seizures (PNES), which are episodes that look like seizures but are driven by psychological rather than electrical processes. Because both frontal lobe epilepsy and PNES can be scalp EEG-negative, telling them apart based on the EEG alone can be impossible.16Neurologic Clinics. Diagnostic evaluation of psychogenic nonepileptic seizures
This overlap is clinically significant because the treatments are completely different. Epileptic seizures are treated with antiseizure medications; PNES are treated with psychological therapy. Misdiagnosis in either direction causes real harm. Most patients who are incorrectly diagnosed with epilepsy turn out to have either psychogenic nonepileptic events or syncope (fainting episodes).17PubMed. The Role of EEG in the Erroneous Diagnosis of Epilepsy Some of these patients spend years on antiseizure drugs they do not need, experiencing side effects for a condition they do not have, because a normal EEG was treated as ambiguous rather than informative.
When Doctors Suspect Epilepsy Despite Normal EEGs
Epilepsy is ultimately a clinical diagnosis. A neurologist can diagnose epilepsy based on a detailed description of the events, witness accounts, and the clinical context, even if every EEG has come back normal. The EEG supports or modifies the diagnosis, but it does not make or break it on its own. If you have had two unprovoked seizures with witnesses who describe typical epileptic features, a normal EEG does not mean you do not have epilepsy. It means the test did not catch the abnormality during the window it was recording.
For patients in this situation, the path forward usually involves a combination of strategies: repeat EEGs with sleep deprivation, extended ambulatory monitoring, and sometimes MRI or other imaging to look for structural brain abnormalities that would support the diagnosis. If the clinical suspicion is very high and the seizures are drug-resistant, intracranial EEG may be considered. This involves surgically placing electrodes directly on or within the brain, bypassing all the signal-dampening layers of skull and scalp. Intracranial monitoring can detect seizures from deep structures and small cortical areas that scalp EEG simply cannot reach.
Magnetoencephalography as an Alternative
Magnetoencephalography (MEG) measures the tiny magnetic fields produced by the same neuronal currents that EEG measures electrically. Because magnetic fields pass through the skull without the same distortion and dampening that electrical signals experience, MEG can detect activity from a smaller patch of cortex. Research has shown that MEG can pick up epileptic spikes arising from as little as 3 to 4 square centimeters of cortex, well below the threshold where scalp EEG goes blind. In one study, spikes detected only by MEG (and missed by simultaneous EEG) were significantly smaller in magnitude than spikes that both technologies could see, confirming that the EEG-invisible spikes were real but simply too small for scalp electrodes to capture.18PubMed Central. Magnetoencephalography’s higher sensitivity to epileptic spikes may elucidate the profile of electroencephalographically-negative epileptic seizures
MEG is not widely available. The equipment is expensive and requires a magnetically shielded room, so it is typically found only at large academic medical centers. It is most commonly used in presurgical evaluation for patients with drug-resistant epilepsy, where precisely localizing the seizure source is critical. For everyday diagnostic work, it remains a niche tool. But its existence is a useful reminder that EEG-negative epilepsy is a limitation of the recording method, not evidence that the seizures are not real or not electrical.
What a Normal EEG Actually Tells You
If you have had a seizure and your EEG came back normal, the honest interpretation is that the test did not detect abnormal activity during the time it was recording. It does not mean your brain is fine, and it does not mean the event was not a seizure. The test has limited sensitivity for the reasons discussed throughout: short recording windows, signal dampening through the skull, deep or small seizure sources, and the intermittent nature of epileptic discharges.
Conversely, an abnormal EEG in someone who has never had a seizure does not automatically mean they have epilepsy. A small percentage of healthy people have mildly abnormal EEG patterns that carry no clinical significance. The EEG is one piece of a diagnostic puzzle that includes the patient’s history, brain imaging, and the clinical judgment of a neurologist. It is a useful tool, sometimes an indispensable one, but it has blind spots large enough that treating a normal result as proof of anything would be a mistake.