Diagnosing epilepsy relies on a combination of clinical history, electrical brain recordings, brain imaging, and careful exclusion of other conditions that can mimic seizures. There is no single test that confirms epilepsy on its own. Instead, the diagnosis comes together like a case file, with each piece of evidence narrowing down what happened in the brain and why.
What Counts as Epilepsy
The formal definition, established by the International League Against Epilepsy, requires meeting one of three conditions. The most straightforward is having at least two unprovoked seizures more than 24 hours apart. But you can also be diagnosed after a single unprovoked seizure if there’s at least a 60% chance of another one occurring within the next 10 years. That threshold is an approximate guideline rather than a hard cutoff; after two unprovoked seizures, the recurrence risk is generally between 60% and 90%. The third pathway is being diagnosed with a recognized epilepsy syndrome, which bundles together specific seizure types, EEG patterns, and clinical features into a known condition.
The word “unprovoked” matters enormously here. A seizure triggered by dangerously low blood sugar, a severe sodium imbalance, alcohol withdrawal, or a high fever in a young child is considered provoked. These seizures happen because of a temporary disruption to the brain, not because of an underlying tendency to seize. If the trigger is removed and no further seizures occur, the person does not have epilepsy.
Epilepsy can also be considered resolved. If someone had a childhood epilepsy syndrome and has aged out of the typical window, or if they’ve been seizure-free for 10 years with at least 5 of those years off medication, the condition is no longer considered active.
Clinical History Is the Foundation
The most important diagnostic tool isn’t a machine. It’s a detailed account of what happened before, during, and after a suspected seizure. Because doctors rarely witness the event themselves, they rely heavily on descriptions from the person and any bystanders who were present.
Several features strongly suggest a true seizure: abrupt onset, unresponsiveness, rhythmic jerking of the limbs, a bitten tongue (especially on the side rather than the tip), loss of bladder control, and confusion afterward. The period after a seizure, called the postictal phase, is particularly informative. Confusion, headache, muscle soreness, and temporary weakness on one side of the body all point toward a seizure and can even help pinpoint where in the brain it started.
Physical clues matter too. A bitten tongue on the side is highly specific to generalized tonic-clonic seizures. Shoulder dislocations, unexplained bruising, and head injuries can corroborate an event that nobody witnessed. Warning symptoms before the seizure, such as a feeling of déjà vu, a rising sensation in the stomach, tingling on one side of the body, or visual disturbances, suggest the seizure started in a specific brain region before spreading.
Ruling Out Conditions That Look Like Seizures
Fainting (syncope) is one of the most common mimics of epilepsy, and distinguishing the two is a core diagnostic principle. Fainting tends to happen while standing or sitting for long periods, in warm environments, or during painful experiences like blood draws. People who faint often feel lightheaded, sweaty, nauseated, or warm beforehand, and they recover quickly without lingering confusion.
Seizures, by contrast, are more likely to involve a bitten tongue, head turning to one side, unusual posturing, blue skin color noticed by bystanders, and significant confusion afterward. Bedwetting during the event, prodromal hallucinations, and muscle pain afterward all tip the scales toward seizure rather than syncope. The pattern of recovery is one of the most reliable distinguishing features: fainting resolves within seconds to a couple of minutes, while postictal confusion after a seizure can last 15 minutes or longer.
Psychogenic nonepileptic seizures (PNES) are another major diagnostic challenge. These episodes look like seizures but aren’t caused by abnormal electrical activity in the brain. Clues that suggest PNES include prolonged episodes, movements that fluctuate or appear out of sync, eyes that stay closed during the event, crying during the episode, and preserved awareness of surroundings. The gold standard for confirming PNES is capturing an episode on video while simultaneously recording brain activity with an EEG, showing that the movements occur without any corresponding electrical seizure pattern.
What the EEG Can and Cannot Tell You
An electroencephalogram records electrical activity across the brain’s surface and is the most important test for supporting an epilepsy diagnosis. It can detect abnormal electrical discharges between seizures, identify whether seizures are coming from one area of the brain or both sides simultaneously, and help classify the specific epilepsy syndrome.
But here’s a fact that surprises many people: a normal EEG does not rule out epilepsy. A single routine EEG, which typically lasts 20 to 30 minutes, catches abnormal activity in only about 25% to 55% of people who genuinely have epilepsy. Up to half of epilepsy patients will have a completely normal result on their first recording, and roughly 10% never show abnormal discharges on any routine EEG.
Sleep deprivation before the test improves these odds significantly. A sleep-deprived EEG has a better yield than a standard recording or one performed under drug-induced sleep, making it the most cost-effective approach for investigating new-onset epilepsy in younger adults. Sleep itself activates certain types of abnormal discharges, particularly in generalized epilepsy syndromes, so the test is often timed to capture a period of drowsiness or light sleep.
When routine recordings aren’t enough, prolonged monitoring in a hospital setting (sometimes over several days) can capture actual seizures on video and EEG simultaneously. This is especially valuable for surgical planning or when the diagnosis remains uncertain.
Brain Imaging to Find the Cause
MRI is the imaging standard for anyone diagnosed with epilepsy, and the quality of the scan matters. The ILAE recommends a specific protocol called HARNESS-MRI, optimized for 3-tesla scanners (stronger than the 1.5-tesla machines still common in many facilities). This protocol uses high-resolution, millimeter-scale images taken in three dimensions, which are critical for spotting subtle structural abnormalities that standard scans miss.
The scan looks for things like areas of abnormal brain development (focal cortical dysplasia), scarring in the hippocampus (often from prolonged febrile seizures in childhood), tumors, vascular malformations, and signs of prior brain injury. Focal cortical dysplasia, one of the most common causes of drug-resistant epilepsy, can be remarkably subtle on imaging. It may appear as a slight blurring of the boundary between gray and white matter, a small patch of abnormal signal, or a funnel-shaped streak extending from the brain’s surface down to the ventricles. These features become invisible when scan resolution is too low, because the blurriness of a low-quality image mimics the very feature the radiologist is looking for.
Computer-aided analysis tools are increasingly used alongside visual inspection, with growing evidence that they can reveal lesions that even experienced radiologists miss on standard review. This is especially true when applied to the high-resolution, multi-contrast images the HARNESS protocol produces.
Classifying the Seizure Type
Diagnosis doesn’t stop at confirming epilepsy. Classifying what type of seizures a person has directly determines treatment. The current framework divides seizures into two main categories: focal seizures, which start in one area of the brain, and generalized seizures, which involve both hemispheres from the outset.
Focal seizures are further described by whether consciousness is affected and what symptoms are present, such as jerking on one side, sensory disturbances, or brief staring episodes. Generalized seizures include the classic convulsive type as well as absence seizures (brief lapses in awareness), myoclonic jerks, and atonic seizures where muscle tone suddenly drops. The ILAE is currently reviewing proposed updates to this classification, including replacing the term “awareness” with “consciousness” and reorganizing some subcategories, though the fundamental focal-versus-generalized distinction remains central.
Beyond seizure type, clinicians look for whether the pattern fits a recognized epilepsy syndrome. Syndromes bundle together a characteristic age of onset, seizure types, EEG findings, and often a known genetic basis. Identifying a syndrome can predict how the condition will behave over time and which treatments are most likely to work.
When Genetic Testing Adds Clarity
Genetic testing has become an important diagnostic tool, particularly in children. It’s most useful when epilepsy occurs alongside developmental delays, intellectual disability, or autism, where the chance of finding a causative genetic change is highest. Epilepsy gene panels identify a cause in roughly 20% of tested individuals, though that number varies depending on age of seizure onset, clinical features, and family history.
For conditions like Lennox-Gastaut syndrome, a severe childhood epilepsy, genetic workup is considered essential. This typically includes chromosomal microarray testing to detect missing or duplicated segments of DNA, followed by broader sequencing if no cause is found on imaging or initial tests. Identifying a specific genetic cause can change management, since certain genetic epilepsies respond to targeted treatments while others are worsened by medications that would otherwise seem like reasonable choices.
In adults with new-onset epilepsy and a clear structural cause on MRI, genetic testing is less commonly needed. But when the cause remains unexplained, particularly in younger adults with a family history of seizures, it can still provide answers.
Putting the Pieces Together
No single test makes the diagnosis. A person with a clear clinical history of two unprovoked seizures, a normal EEG, and a normal MRI still has epilepsy. Conversely, someone with abnormal spikes on an EEG but no clinical seizures does not. The EEG supports the diagnosis, imaging looks for a cause, and genetic testing may explain the underlying mechanism, but the clinical story remains the anchor. Each layer of testing refines the picture: what type of epilepsy it is, where in the brain it originates, what’s causing it, and how it’s likely to behave over time. That full picture is what guides treatment decisions and helps predict long-term outcomes.