Ambulatory EEG is a portable, at-home recording of brain electrical activity that usually lasts 24 to 72 hours and is designed to capture events that a routine 20- to 30-minute EEG in a clinic is likely to miss. A technician applies scalp electrodes in the office, the patient goes home with a small recorder, and a neurologist later reads the study against a simultaneous video and event diary. It is most useful for distinguishing epileptic seizures from non-epileptic events, quantifying how often discharges occur, and evaluating spells that happen mainly during sleep or at unpredictable times.
What the electrodes actually record
Every scalp electrode measures a voltage difference between two points on the head, sampled at 200 to 512 times per second in a typical clinical setup. The signals that matter are in the 1 to 30 Hertz range, the same frequencies as the rhythmic firing of cortical neuron populations. A seizure produces abnormally synchronized firing across a patch of cortex, which appears as a sudden change in frequency, amplitude, or morphology on the tracing.
The limitation is anatomy. Scalp electrodes sit several millimeters from the brain, separated by scalp, skull, and cerebrospinal fluid, all of which blur and weaken the signal. A discharge must involve a reasonably large area of cortex — roughly 6 to 10 square centimeters in most estimates — before it becomes visible on the surface. Deep sources such as mesial temporal structures may generate a clear seizure that barely registers on the scalp.
In plain terms: ambulatory EEG listens through a wall, so it hears loud events well and quiet ones poorly.
That physics explains why the test is judged on whether it captured the event, not on whether the brain looked normal. A normal-looking 48-hour recording carries real information only if the patient actually had a spell during it.
Timeline: why 24 hours is often not enough
The single most important number in ambulatory EEG is the yield curve, the probability of capturing a typical event as recording time increases. Yield depends on how often the patient's spells occur, so a daily spell and a monthly spell are entirely different problems.
| Recording duration | Approximate chance of capturing a typical event | Practical notes |
|---|---|---|
| 20-30 minutes (routine EEG) | 1 to 3 percent in patients with infrequent spells | Captures the event only if it happens during the visit |
| 6 hours | Highly variable; useful for very frequent or sleep-predominant spells | Often used in children with daily events |
| 24 hours | Roughly 20 to 50 percent when spells occur several times per week | First-line home duration in many centers |
| 48 hours | Meaningfully higher than 24 hours for weekly spells | Common compromise for adults |
| 72 hours | Near the practical ceiling for most home studies | Skin irritation and compliance begin to limit value |
| 5 to 7 days | Used selectively for rare or unpredictable events | Requires careful electrode maintenance |
| Inpatient video-EEG (days) | Highest yield, with continuous supervision | Reserved for presurgical workup or diagnostic dilemmas |
The pattern is arithmetic, not magic. If a patient has one spell per week, a 48-hour recording covers about 29 percent of a week, so the chance of it falling inside the window is roughly that order of magnitude — higher if spells cluster, lower if they are unpredictable. Spells that occur mainly in sleep argue for overnight inclusion, which is why a 24-hour study that starts in the morning captures one night.
The mechanism behind the yield curve is simple: capture requires coincidence between event occurrence and recording time. Nothing about the machine improves with duration except the size of the window. This is why no guideline sets a single "correct" duration — the duration should match the event frequency.
What ambulatory EEG can and cannot replace
The main alternative is inpatient video-EEG monitoring, typically 3 to 7 days in a hospital bed with continuous technologist supervision and synchronized video. Ambulatory systems also record video in many modern setups, but with a fixed camera, no supervision, and no ability to taper anti-seizure medication under observation unless a physician designs that plan.
| Feature | Ambulatory EEG | Inpatient video-EEG | Routine EEG |
|---|---|---|---|
| Typical duration | 24 to 72 hours | 3 to 7 days | 20 to 30 minutes |
| Supervision | None or remote review | Continuous technologist | Technologist present |
| Video quality | Fixed, limited angles | Multi-angle, staff-adjusted | None |
| Medication taper | Rarely | Common | No |
| Cost (US, approximate) | USD 1,000 to 3,000 | USD 5,000 to 20,000+ | USD 200 to 800 |
| Best for | Frequent spells, sleep events, quantification | Rare spells, presurgical workup, diagnostic uncertainty | Screening, first assessment |
| Main weakness | Event may not occur; artifact | Cost, access, disruption | Very low yield for infrequent events |
The cost figures are order-of-magnitude and vary widely by region and payer. The clinical distinction matters more: ambulatory EEG is a capture device, while inpatient monitoring is a diagnostic environment that allows provocation and medication manipulation.
A corrected misconception worth stating plainly: ambulatory EEG is not a cheaper substitute for inpatient monitoring in a patient with rare, medically refractory spells being considered for surgery. In that setting, the events may not happen at home, and the recording lacks the supervision needed to interpret subtle behavioral features.
Artifacts and non-epileptic events
Roughly 20 to 40 percent of patients referred for spell evaluation turn out to have non-epileptic events, and ambulatory EEG is often the test that sorts this out. The catch is that home recording is artifact-rich: chewing, eye movements, pulse, sweating, electrode pops, and phone use all produce signals that can mimic discharges.
Experienced readers use several checks. True cortical discharges usually appear across multiple adjacent electrodes in a physiologically plausible field. Artifacts tend to be isolated, sharply contoured, and time-locked to a movement or a diary entry. Simultaneous video and a patient event button are the strongest tools, which is why a diary with timestamps is not optional — it is part of the test.
In plain terms: the diary tells the reader where to look, and the video tells them whether the body did what the brain tracing suggests.
A non-obvious caveat: a normal ambulatory EEG does not exclude epilepsy. If no event occurred during the recording, the study is indeterminate, not negative. Reporting it as "normal" is a common and consequential error.
Seizure frequency, sleep, and who benefits most
The patients who benefit most are those with events frequent enough to be captured in 24 to 72 hours, typically several per week or at least several per month. Patients with one spell per year are poor candidates for home recording; their expected capture probability in 48 hours is under 1 percent.
Sleep is a major variable. Many focal epilepsies, and several generalized syndromes, activate during non-REM sleep. A recording that includes at least one full night captures the period of highest discharge density for those patients. This is one reason 24-hour studies starting in the morning outperform same-day shorter studies in yield for sleep-predominant events.
Children present a different profile. Pediatric absences and some focal epilepsies occur many times daily, so a 6- to 24-hour study often suffices. Behavioral tolerance of electrodes is the limiting factor, and parent-reported diary quality drives interpretation.
Quantification is an underappreciated use. Counting interictal discharges per hour, or measuring how discharge density changes after a medication adjustment, gives a number that a routine EEG cannot. A drop from, say, 40 discharges per hour to 5 after a drug change is a measurable signal, though the link between discharge density and seizure control is statistical, not deterministic.
Limitations and the honest edge cases
Ambulatory EEG has real limits that no amount of duration fixes. Scalp signal is blind to small or deep sources, so a normal study in a patient with convincing focal spells does not rule out a deep-seated lesion. Electrode impedance rises over days as skin oils and sweat accumulate, degrading signal quality after roughly 48 to 72 hours without maintenance.
Skin reactions are common, with irritation reported in a meaningful minority of patients beyond 72 hours. Hair washing is restricted during the study, which affects compliance.
The interpretation burden is also real. A 72-hour recording generates a large volume of data, and review time per study is substantial. Some centers use automated spike-detection software, but human review remains the standard, and automated tools have meaningful false-positive rates that require confirmation.
Established science versus hypothesis should be kept separate here. That ambulatory EEG captures events and distinguishes epileptic from non-epileptic spells is well established. That specific discharge counts predict individual seizure risk is a working hypothesis supported by group data, not a reliable individual forecast.
What to actually do before, during, and after the study
Preparation determines whether the recording is interpretable. Concrete actions:
- Shampoo hair the night before and skip conditioners, oils, and styling products, which raise electrode impedance.
- Bring a written list of current medications, doses, and times, and confirm with the ordering physician whether any should be held.
- Plan a normal day, including work or school if safe, because provoking typical triggers improves capture odds.
- Set up the event button and diary in advance and rehearse pressing it the moment a spell begins.
- Arrange a second person to note the time, behavior, and duration if consciousness is impaired.
- Charge or plug in the recorder per the manufacturer's instructions and keep it dry.
During the study, log every event with a timestamp, even if it seems minor. Note sleep and wake times. Avoid swimming, bathing, and hair products.
Afterward, return the equipment on schedule and review the diary with the reading neurologist so the events are matched to the tracing. Ask specifically whether a typical event was captured. If the answer is no, the study is incomplete, and repeating it with a longer duration or moving to inpatient monitoring is a reasonable next step.
Red flags that change the plan
Some findings or situations should override a routine ambulatory plan and prompt urgent or supervised evaluation:
- A first-ever event with prolonged confusion, focal weakness, or incomplete recovery.
- Events with injury, aspiration, or breathing difficulty.
- Suspected status epilepticus or prolonged seizures beyond 5 minutes.
- New focal neurological deficit, fever with headache, or signs of raised intracranial pressure.
- Pregnancy with new seizures, or any event in a patient on anticoagulation with head trauma.
- Rapidly increasing event frequency or a change in event character.
These are not ambulatory EEG problems; they are emergencies or inpatient questions.
Key numbers reconciled
Thresholds in this article are consistent if read together: the 20- to 30-minute routine EEG captures only 1 to 3 percent of infrequent events, a 24- to 72-hour ambulatory study raises the capture window to roughly a fifth to half of patients with several weekly spells, and durations beyond 72 hours are limited by impedance rise and skin irritation rather than by diagnostic value. The 5-minute status epilepticus threshold and the 48- to 72-hour electrode maintenance window are different scales serving different purposes, not competing numbers.