Theta Waves: Meaning, Meditation and Evidence

What are theta waves?

Theta waves are rises and falls in the brain's electrical activity that repeat 4 to under 8 times a second. The glossary of the International Federation of Clinical Neurophysiology (IFCN) sets the theta band at 4 to under 8 Hz, and a single theta wave at 125 to 250 milliseconds long.

They are recorded with EEG, electroencephalography: electrodes on the scalp pick up the summed activity of millions of neurons beneath. Theta sits between two better-known rhythms. Below it is delta, under 4 Hz, the slow waves of deep sleep. Above it is alpha, 8 to 13 Hz, which the glossary describes as best seen "with the eyes closed, during physical relaxation and relative mental inactivity". Beta, above 13 to 30 Hz in the glossary's correction, is faster again.

The bands and how long one wave lasts

Band	Range	One wave lasts
Delta	0.5 to under 4 Hz	250 ms to 2 s
Theta	4 to under 8 Hz	125 to 250 ms
Alpha	8 to 13 Hz	77 to 125 ms
Beta	above 13 to 30 Hz	33 to 77 ms

The names are labels on a continuum, not switches in the brain. Critics since 1938 have seen the EEG spectrum as "a continuum", as Edwards and Chang's history of the spectrum puts it, and authors have drawn the edges of the bands in slightly different places. The guide to brainwave frequencies takes every band in turn, and theta's neighbours, delta waves and alpha waves, have guides of their own.

Where does the name theta come from?

From drowsiness. In 1941 the German researcher R. Jung described an "intermediate δ" band of about 5 to 7 Hz as a sign of drowsiness, and Edwards and Chang describe it as the precursor of theta. The theta band as it is used today goes back to a 1944 paper by W. Grey Walter and V. J. Dovey on brain recordings in people with tumours beneath the cortex, published in the Journal of Neurology, Neurosurgery and Psychiatry. Walter had already introduced delta in 1936, for slow rhythms recorded near tumours. Theta, the Greek letter θ, took the gap between delta and alpha.

When does the brain make theta waves?

Most clearly as you drift off. During "the transition from waking to sleeping", Klimesch's 1999 review found, "upper alpha power decreases, whereas theta increases". The change is gradual. Ogilvie's review of falling asleep concluded that sleep does not begin "in a moment". It is a series of changes from relaxed drowsiness through the first stage of sleep into the second, the early ones tied to alpha and the later ones to the growth of theta.

Harmonance measured that change in open recordings. The Sleep-EDF database on PhysioNet holds 153 nights recorded at home between 1987 and 1991 from healthy adults aged 25 to 101, scored by trained technicians in 30-second pages (Kemp and colleagues). Harmonance lined up 152 of those nights, from 77 people, on their first 30 seconds scored as sleep. For each 30 seconds in the ten minutes either side, it measured how the brain's electrical power from 0.5 to 30 Hz was shared between the bands, over the back of the head.

Theta and alpha either side of falling asleep

When	Theta share	Alpha share
10 minutes before	9.6%	13.1%
5 minutes before	10.4%	16.8%
1.5 minutes before	12.3%	26.3%
First 30 seconds asleep	15.5%	17.6%
5 minutes after	14.7%	9.4%
9.5 minutes after	16.6%	9.8%

Download the sleep onset data (CSV)

Theta's share rose from 9.6% ten minutes before sleep to 15.5% in the first 30 seconds scored as sleep, and stood at 16.6% nine and a half minutes later. Alpha, the eyes-closed rhythm, did the opposite after a last rise: it peaked at 26.3% a minute and a half before sleep, then fell to 9.8%.

Two things keep that reading honest. First, theta's own power hardly moved at the moment of falling asleep, 4.5 square microvolts ten minutes before and 3.8 in the first 30 seconds of sleep, so its share jumped mostly because alpha fell away; theta's power then climbed to 6.8 over the next ten minutes. Second, scorers mark the first stage of sleep partly by these same changes, the "low amplitude mixed frequency EEG activity" of light sleep in the IFCN glossary. So the chart shows what falling asleep looks like in the recordings. It does not show that theta causes sleep, or that adding theta to your ears would. Falling asleep is not one switch either: in the ten minutes after that first sleep epoch, only 70 to 81% of nights were still scored asleep at any given minute.

Theta also comes with hard thinking. The glossary lists a theta rhythm over the middle of the forehead that "occurs during psychological stress and cognitive tasks, particularly problem solving", seen mostly in adults under 30. Cavanagh and Frank argue that this frontal theta marks the moment the brain registers a need for control, and that it reacts to "novelty, conflict, punishment and error".

Much of what is written about theta comes from rats. When a rat runs through its surroundings, its hippocampus, a structure deep in the brain studied for its part in memory and navigation, shows large, regular theta waves. Buzsáki called them the "on-line" state of the hippocampus. In people, direct recordings from inside the brain point to a slower matching rhythm of about 1 to 4 Hz, as Jacobs reviews. The theta on a scalp recording and the hippocampal theta of animal studies are not the same thing.

Is more theta better?

It depends on when. Klimesch's review links good memory performance to two patterns that look like opposites: less theta at rest, and a burst of theta during the task itself. Resting theta also rises after long spells without sleep. Edwards and Chang draw the same line between spontaneous theta, "generally a sign of drowsiness and idling", and theta driven by a task. A high level of theta at rest is not a sign of a sharper mind.

What happens to theta waves in meditation?

They often grow. A systematic review of 56 EEG studies of mindfulness, with 1,715 people, found meditation most often linked with more alpha and more theta than resting with the eyes closed, "although such outcomes were not uniformly reported". An earlier review by Cahn and Polich described "an overall slowing" of the EEG with meditation, with theta and alpha tied to how practised the meditator was. In one small study of experienced meditators, the state they rated as blissful came with more theta at the front and midline of the head (Aftanas and Golocheikine).

The direction runs from practice to theta. These studies show what the brain does during meditation. They do not show that adding a theta sound brings a meditative state, which is the question of the next section. If you like to sit with sound, the Falling Theta Session and the reflective page of How do you feel? gather tones people describe reaching for.

Can a sound bring on theta waves?

A binaural beat is the usual attempt. Play a steady tone in each ear, slightly apart in pitch, and your hearing makes a beat at the difference: 108 Hz in the left ear and 115 Hz in the right give a beat of 7 a second, even though "no physical combination of the tones occurs outside the auditory system" (López-Caballero and Escera). Binaural beats are heard from about 1 to 30 Hz, "a range that coincides with the main human EEG frequency bands", and that overlap is where the idea of steering brain rhythms with them comes from (Ingendoh and colleagues).

The two tones at the start

Ear	Tone	Waves in one beat
Left	108 Hz	15.43
Right	115 Hz	16.43
Difference	7 Hz	1

Whether the brain follows is unsettled. Ingendoh and colleagues' systematic review of 14 EEG studies counted five that supported the idea, eight that contradicted it and one with mixed results, and found the studies very varied in how they played the beats and what they measured. For theta beats in particular, the results split the same way.

Theta binaural beat studies

Study	What was played	What was found
Brady and Stevens, 2000	a theta protocol, three sessions of 20 minutes, 6 people	theta rose in 5 of 6
Stevens and colleagues, 2003	theta beats, 4 hours in all, double-blind	no rise in frontal theta
Wahbeh and colleagues, 2007	7 Hz (133 and 140 Hz), 30 minutes, 4 people	no EEG difference; lower immediate recall
Jirakit­tayakorn and Wongsawat, 2017	6 Hz on a 250 Hz tone, 30 minutes, 28 people	theta rose within 10 minutes
López-Caballero and Escera, 2017	4.53 Hz among five beats, 3 minutes each	no change in the matching band
Garcia-Argibay and colleagues, 2019	5 Hz before a memory test, 32 people	fewer words recalled
Lane and colleagues, 1998	1.5 and 4 Hz in a 30-minute watch task, 29 people	fewer correct detections than beta beats
Ingendoh and colleagues, 2023	a review of 14 EEG studies, all bands	5 for, 8 against, 1 mixed

Brady and Stevens played six people a theta protocol and saw theta rise in five. A larger, double-blind follow-up by Stevens and colleagues, with 4 hours of beats, found no rise in frontal theta and called the effect "a very low power phenomenon". Wahbeh and colleagues played the session's own starting beat, 7 Hz (133 and 140 Hz), to four adults for 30 minutes and found no difference in any EEG measure. Jirakittayakorn and Wongsawat found theta rising across the scalp within 10 minutes of a 6 Hz beat in 28 people, while López-Caballero and Escera saw no change with a 4.53 Hz beat played for 3 minutes.

Do theta binaural beats have a downside?

For anything that needs memory or attention, possibly. Three small studies point the same way. After 30 minutes of the 7 Hz beat, Wahbeh's four adults recalled fewer words straight afterwards, and the authors asked for larger studies to check it. In 32 people, a 5 Hz theta beat before a memory test "lessened the number of correctly remembered words", while a 20 Hz beta beat went with more (Garcia-Argibay, Santed and Reales). And in 29 people on a 30-minute watch task, beta beats went with more correct detections and less negative mood than theta and delta beats of 1.5 and 4 Hz (Lane and colleagues).

So if you want sound while you study or work, the evidence points away from theta. Across 22 studies, a 2019 meta-analysis found a medium average effect of binaural beats on tests of memory and attention and other measures, and found that "the direction and the magnitude of the effect depends upon the frequency used, time under exposure, and the moment in which the exposure takes place".

What is the Falling Theta Session?

Nine hours of two pure tones. Your left ear hears 108 Hz throughout. Your right ear hears 115 Hz gliding evenly down to 113 Hz, so the beat between them slows from 7 to 5 a second and stays inside theta the whole time. It plays free, with no account, on its own page.

Harmonance measured the nine-hour file uploaded for the long video, not just the plan. Across 6,481 measuring windows, neither tone strays more than 0.0003 Hz from where it should be. The left ear reads 108.0002 Hz at the start and 108.0000 Hz at the end; the right ear reads 115.0000 Hz at the start, 113.9999 Hz halfway and 113.0001 Hz at the end. Each ear's tone sits at least 114 dB above any trace of the other ear's, so on headphones the beat can only form inside your own hearing. The file is mastered quietly, its loudest moment 21 dB below the digital maximum.

The arithmetic is worth doing once. Over nine hours your right ear hears 3,693,600 waves and your left 3,499,200: the difference is 194,400 beats. One beat lasts 143 milliseconds at the start and 200 at the end, the same span as a single theta wave in the IFCN glossary, 125 to 250 milliseconds. The right ear's tone falls by 30.4 cents, under a third of a semitone, and the beat slows by 0.22 a second every hour.

The beat, hour by hour

Hour	Beat	One beat lasts
Hour 1	7 Hz	143 ms
Hour 2	6.778 Hz	148 ms
Hour 3	6.556 Hz	153 ms
Hour 4	6.333 Hz	158 ms
Hour 5	6.111 Hz	164 ms
Hour 6	5.889 Hz	170 ms
Hour 7	5.667 Hz	176 ms
Hour 8	5.444 Hz	184 ms
Hour 9	5.222 Hz	191 ms
End of hour 9	5 Hz	200 ms

Its neighbours in the Binaural Beats collection are the Falling Delta Session, which slows from 3.5 to 1.5 Hz, and the Falling Alpha Session, from 12 to 9 Hz. Steady theta beats play at 6 Hz and 4 Hz. Do binaural beats work? weighs the evidence across every band.

How long should you listen to binaural beats?

As long as it stays comfortable; studies have used much less than nine hours. The theta studies above played beats for 3 minutes, for 20 or 30 minutes, and for 4 hours spread across sessions, and the 2019 meta-analysis found longer listening did better on its measures. No study has played a beat for nine hours, and none has tested this session. Its length is there so you can start anywhere and stop whenever you choose: a meditation or a quiet half hour needs only a slice of it.

How should you listen?

Use headphones. The beat only forms when each ear hears its own tone; on speakers the two tones meet in the air and make an ordinary loudness beat instead. Start with the volume low, lower than you think, and stop if anything feels uncomfortable. Do not listen while driving or doing anything that needs your full attention, since in one study theta and delta beats went with fewer correct detections on a watch task. Then press play and notice what you notice.

Sources and limits

Every outside source was opened and its words checked on the day this guide was written. Each one is linked where it is used above.

• Definitions of the bands and waves: the IFCN glossary of EEG terms (Kane and colleagues, 2017) and its correction (2019).
• The history of the name: Edwards and Chang (2013) and Walter and Dovey (1944).
• Theta and falling asleep: Klimesch (1999) and Ogilvie (2001), both reviews.
• The sleep onset chart: Harmonance's own analysis of the Sleep-EDF Database Expanded (Kemp and colleagues, 2000), open under the Open Data Commons Attribution licence. Its limits: home recordings from 1987 to 1991, scored with the 1968 Rechtschaffen and Kales rules, from electrodes at Fpz-Cz and Pz-Oz rather than the positions used today; one of the 153 nights is left out because its recording is stored in 60-second blocks; and medians hide large differences between people, which the shaded band shows.
• Frontal and hippocampal theta: Cavanagh and Frank (2014), Buzsáki (2002) and Jacobs (2014).
• Meditation: Lomas and colleagues (2015), Cahn and Polich (2006) and Aftanas and Golocheikine (2001).
• Binaural beats: Brady and Stevens (2000), Stevens and colleagues (2003), Wahbeh and colleagues (2007), Jirakittayakorn and Wongsawat (2017), López-Caballero and Escera (2017), Garcia-Argibay, Santed and Reales (2019), Lane and colleagues (1998), the 2019 meta-analysis and the 2023 systematic review. Most are small, and they differ in tones, beat rates and listening times; none played a beat for nine hours.
• The session's numbers: Harmonance's own measurement of the uploaded nine-hour file, and arithmetic from its two tones.