Japanese GABA & Glycine Supplements: What Glycine Sleep Studies Show
|Japan Pantry
Does glycine help people fall asleep faster, feel better about their sleep, or feel less tired the next day? Those are three different questions. Small human trials from Japanese research teams have investigated all three, but a favorable result on one measure does not establish an improvement on every measure—or an effect from every supplement that contains glycine.
1. The short answer: promising signals, limited certainty
Quick verdict: Early Japanese trials reported favorable findings for some subjective sleep ratings, some laboratory sleep-onset measures, and selected next-day outcomes. The evidence remains preliminary: the trials were small, brief, and included industry-affiliated researchers. A systematic review published online in 2023 described the healthy-population sleep studies as having small sample sizes and a high risk of bias. [1][2][3][4]
The useful question is not simply “Does glycine work?” It is which outcome changed, in whom, under what conditions, and how confidently can that finding be applied elsewhere? This article examines that evidence rather than ranking products by sleep claims. Our separate Japanese GABA buying guide compares GABA-containing products by their labels.
The three trials discussed below used 3 g of glycine before bed. That is a research regimen, not a personal dosing recommendation. A retail formula containing glycine plus GABA or other ingredients is not the same intervention.
Three questions that should remain separate
A person might feel more refreshed without a measurable change in the distribution of sleep stages. Another might reach a particular stage sooner without spending more of the night in that stage. A next-day task might show faster responses even when the participant's sleepiness rating does not change significantly. These are not necessarily contradictions: they are different outcomes, collected at different times, with different sources of uncertainty.
This distinction also explains why a headline such as “glycine improves sleep” can be too broad for the underlying evidence. Before applying it to a shopping decision, identify whether the headline refers to a questionnaire, an overnight recording, or a daytime test. The Japanese studies provide useful examples precisely because they did not all measure the same thing.
2. “Better sleep” is not one measurement
Sleep research can ask people how they felt, record their sleep in a laboratory, or test how they functioned afterward. Each approach answers a different question.
Swipe to compare →
Outcome
What researchers measured
What it does not establish by itself
Subjective sleep quality
Questionnaire ratings of the previous night and the feeling on waking.
A measured increase in deep sleep or total sleep time. [1]
Sleep-onset latency
The time until a defined sleep stage appears on polysomnography, or PSG.
A better result on every sleep-quality rating or a lasting benefit outside that laboratory. [2]
Slow-wave-sleep latency
The time until slow-wave sleep first appears.
More total deep sleep. Earlier arrival and a greater amount are different outcomes. [2]
Next-day fatigue
A person's rating of tiredness, including visual analog scales.
The same change as reduced sleepiness or better performance on all cognitive tasks. [3]
Next-day performance
Specific tasks, such as psychomotor vigilance or memory recognition.
A general improvement in concentration, productivity, or driving safety. [2][3]
Subjective outcomes are meaningful: feeling more rested matters. They are simply not interchangeable with instrument-based outcomes. Likewise, a statistically significant finding does not tell us how noticeable or durable the difference would be for an individual.
What PSG adds—and what it does not
Polysomnography records physiological signals during sleep, allowing researchers to identify sleep stages rather than relying only on a person's estimate. In the 2007 trial, sleep onset was defined by the first appearance of stage 2 sleep. A different study using a different definition could produce a different latency even if the night's underlying pattern were similar. When comparing studies, read the definition alongside the number. [2]
“Sleep efficiency” also needs a denominator: it describes sleep time relative to time in bed. In the 2007 paper, the favorable subjective sleep-efficacy result came from the sleep questionnaire. It should not be silently converted into an instrument-confirmed increase in efficiency. Likewise, time in bed in the 2012 restriction experiment was tracked using sleep logs, not equivalent to a full-night PSG assessment of every sleep stage. [2][3]
Why a crossover trial helps, but does not settle the question
In a crossover design, the same participants receive both interventions in separate periods. This can reduce the influence of stable differences between people, since a participant is compared with their own response under the other condition. It does not eliminate expectations, changes between nights, carryover effects, missing data, or chance findings across many outcomes. Washout periods and blinding therefore matter alongside sample size.
Repeated measurements do not create additional participants. Seven people assessed three times in a day are still seven people, not a 21-person trial. That is particularly important when reading the graphs below: each pair of bars represents another assessment of the same small crossover sample.
3. What the Japanese human trials actually found
Japan's early glycine research is directly relevant to this topic. Looking at the original studies is more informative than repeating a manufacturer's general description of “rest” or “refreshing mornings.”
Swipe to compare →
Study
Participants and design
Intervention and duration
Main interpretation
Inagawa et al., 2006
19 enrolled; 15 completed and contributed to the efficacy evaluation. Randomized, double-blind crossover in volunteers dissatisfied with their sleep.
3 g before bed; four nights per treatment period, with a three-day washout.
Favorable findings on selected next-morning subjective ratings, including fatigue, liveliness, and clear-headedness. This was not a PSG study. [1]
Yamadera et al., 2007
11 healthy volunteers, eight women and three men. Randomized, single-blind crossover with overnight PSG and daytime testing.
3 g within one hour before bed; two consecutive nights per treatment period, separated by at least one week.
Selected subjective measures and PSG sleep-onset timings improved; the proportions of the different sleep stages did not change. [2]
Bannai et al., 2012
10 healthy men enrolled; seven were analyzed after one dropout and two exclusions. Randomized, single-blind crossover.
3 g 30 minutes before bed during three nights with time in bed restricted by 25%.
Day-one fatigue ratings and selected vigilance outcomes favored glycine. Sleepiness ratings and other measures did not show the same consistent pattern. [3]
The 2006 study: how people felt the next morning
Inagawa and colleagues used the St. Mary's Hospital Sleep Questionnaire and a fatigue checklist. Their findings included favorable ratings for fatigue, liveliness and peppiness, and clear-headedness. Because the outcomes were self-reported, the study cannot establish that participants obtained more deep sleep. The short treatment periods and the loss of four enrolled participants also limit how much can be inferred. [1]
The 2007 study: earlier sleep stages, not more deep sleep
Yamadera and colleagues added PSG. Glycine shortened the time until stage 2 sleep—the study's sleep-onset measure—and the time until the first appearance of stage 3 sleep. However, REM-sleep latency did not change, and sleep architecture, expressed as the proportions of time in the different stages, was not altered. “Reached slow-wave sleep sooner” is therefore more accurate than “increased deep sleep.” The study also reported selected favorable subjective and daytime-task results, but it involved only 11 volunteers. [2]
Swipe to compare →
2007 outcome
Reported result
Careful interpretation
PSG latency to stage 2
Shorter with glycine; p = 0.010.
Earlier sleep onset under this study's definition, in a small laboratory trial.
PSG latency to slow-wave sleep
Shorter with glycine; p = 0.019.
Earlier first appearance of stage 3, not evidence of more total deep sleep.
REM-sleep latency
Not changed.
The latency finding did not extend to every sleep stage.
Sleep-stage proportions
Sleep architecture was not altered.
Do not describe this as a redistribution of the night toward deeper sleep.
Subjective sleep efficacy
A favorable questionnaire result.
A self-reported sleep-time/in-bed-time measure, distinct from a PSG-derived outcome.
These results are reported from the original paper; they are not estimates read from a marketing image. A p-value describes the result of a statistical test under its assumptions. It is neither the size of the benefit nor the probability that a particular customer will notice a change. The small sample and short observation period remain important even when a test crosses the conventional significance threshold. [2]
The 2012 study: fatigue is not the same as sleepiness
Bannai and colleagues tested a different situation: healthy men whose usual time in bed was deliberately shortened, from about 7.3 to 5.5 hours. On day one, the treatment effect on the fatigue visual analog scale was statistically significant (p = 0.022), while the corresponding sleepiness result was not (p = 0.098). On day three, neither of those subjective outcomes showed a significant difference. The fatigue-complaints questionnaire also did not show significant treatment differences on day one. [3]
Psychomotor-vigilance reaction time favored glycine on days one and three, but the other performance tests did not show significant differences. These results do not establish that glycine replaces adequate sleep or improves every aspect of daytime functioning. With only seven men in the analysis, the findings need cautious interpretation and stronger replication. [3]
The actual day-one numbers: fatigue versus sleepiness
The following table reproduces the means and standard errors from the study's Table 1. Both visual analog scales ran from 0 to 100, with higher scores representing more fatigue or sleepiness. The final column is a calculated difference between the two reported means, not a new statistical analysis. [3]
Swipe to compare →
Day-one outcome
Assessment time
Placebo: mean (SE)
Glycine: mean (SE)
Glycine minus placebo: score points
Fatigue (0–100)
10:00
48.3 (4.97)
33.6 (4.60)
-14.7
Fatigue (0–100)
14:00
53.1 (6.20)
35.4 (8.62)
-17.7
Fatigue (0–100)
18:00
59.6 (8.14)
51.0 (9.73)
-8.6
Sleepiness (0–100)
10:00
48.1 (5.65)
32.7 (7.94)
-15.4
Sleepiness (0–100)
14:00
52.9 (10.51)
36.0 (6.78)
-16.9
Sleepiness (0–100)
18:00
51.0 (6.97)
45.6 (8.71)
-5.4
Graph 1: Day-one fatigue ratings. Replotted from Bannai et al. (2012), Table 1. Bars show means on the full 0–100 scale; standard errors appear in the table above, not as graph error bars. The day-one treatment effect was significant (p = 0.022). This does not establish that every assessment time or every participant improved. Tap the graph for full size. [3]Graph 2: Day-one sleepiness ratings. The glycine means were numerically lower, but the treatment test was not statistically significant (p = 0.098). A visual difference between bars is not enough to establish an effect. The graph uses the same scale, sample and assessment times as Graph 1. Tap the graph for full size. [3]
The contrast between these graphs is the central point: two outcomes can look similar visually yet lead to different statistical conclusions. The paper also reported a significant post-hoc fatigue comparison at 14:00. That does not justify adding significance marks to every fatigue bar pair. Nor does the nonsignificant sleepiness result prove that the true effect is exactly zero; it means this small experiment did not establish that outcome convincingly. [3]
Standard error describes uncertainty around an estimated mean; it is not the range of individual participants' scores. The plots omit error bars to keep the two outcomes readable, but the numeric table retains the reported standard errors. Neither graph shows a long-term effect, an effect in women, or a tested effect of the retail formulas linked below.
The daytime performance findings also require an outcome-specific reading. For psychomotor-vigilance reaction time, the reported treatment p-values were 0.003 on day one and 0.050 on day three; correct-response results were not significant. For memory-recognition accuracy, the day-three treatment main effect was not significant, although the paper's table reported a treatment-by-time interaction of 0.044. That interaction is not equivalent to an overall improvement in memory. The findings should not be simplified into a general claim of better cognitive performance. [3]
4. Japanese mechanism research—and why the review still urges caution
A proposed mechanism is a separate kind of evidence
A Japanese-linked research team also investigated how glycine might influence sleep and body temperature. Kawai and colleagues' paper, published online in December 2014 and in the 2015 volume of Neuropsychopharmacology, studied rats. It examined peripheral blood flow, heat loss and NMDA-receptor activity in the suprachiasmatic nucleus. It did not provide another human trial of the three retail products in this article. [8]
The experiments supported a proposed connection between glycine, heat dissipation and sleep in that animal model. This is useful biological context, but the route of administration, experimental manipulations and species limit direct translation. A plausible pathway does not tell a shopper whether a particular blend will produce a noticeable benefit, how long any change will last, or what amount is suitable for them.
Swipe to compare →
Evidence layer
What it contributes
What must remain separate
Japanese human trials
Measurements of subjective sleep, PSG timings or next-day outcomes in small samples.
Effects on a different population or a different retail formulation. [1][2][3]
Kawai et al., 2015: rats
A proposed NMDA-receptor/heat-loss mechanism and experimentally observed sleep changes.
Proof that the same pathway explains a human benefit from an ordinary product serving. [8]
Animal experiments in the 2012 paper
Investigation of melatonin and circadian-related gene expression; plasma melatonin was not significantly changed.
A claim that glycine increases melatonin in humans or universally resets their body clock. [3]
Manufacturer label information
The serving, ingredients and quantities in a specific stocked product.
Clinical proof of a sleep or daytime-performance outcome. [5][6][7]
What the systematic review can—and cannot—settle
Soh and colleagues reviewed glycine administration across multiple physiological systems, not just sleep. Their systematic review searched four databases through April 29, 2022. It was published online on October 18, 2023, and appeared in the 2024 volume of GeroScience. Those dates describe different stages of publication; they do not mean the search included every study published through 2023 or 2024. [4]
The review recognized favorable sleep findings in healthy participants but emphasized small sample sizes and a high risk of bias. That is a reason to treat the early results as signals to investigate, not as settled evidence of a reliable benefit for all users.
Small samples: An analysis of seven or 11 people can give unstable estimates and may not represent a wider population.
Brief exposure: A few nights do not establish sustained benefit or long-term safety.
Different settings: Volunteers dissatisfied with sleep and healthy men under temporary sleep restriction are not interchangeable populations.
Mixed outcomes: Favorable results coexist with outcomes that did not reach statistical significance.
Researcher affiliations: The early studies included Ajinomoto-affiliated researchers. In the 2012 trial, participants were Ajinomoto employees. This does not automatically invalidate a result, but independent replication matters. [1][2][3]
Japanese research provides useful context for a store shipping products from Japan. It is not, by itself, proof that a Japanese retail product reproduces the trial results.
Swipe to compare →
Uncertainty
Why it matters here
What stronger evidence would add
Sample size and exclusions
Only seven participants remained in the 2012 analysis; exclusions can affect both precision and representativeness.
A larger, prospectively planned sample and transparent handling of missing data.
Blinding and expectations
Two of the three trials were single-blind; subjective outcomes are particularly sensitive to expectations.
Well-described double-blinding, including how comparable the interventions were.
Many outcomes and time points
A favorable result on one test can coexist with null results elsewhere.
Prespecified primary outcomes and an analysis plan that addresses multiple comparisons.
Short treatment periods
A few nights cannot establish a durable effect or long-term safety.
Longer follow-up with both benefits and adverse events systematically assessed.
Affiliations and population
Industry involvement and an employee sample make independent confirmation especially useful.
Independent replication in a broader sample rather than relying on the same research network.
A high risk-of-bias assessment is not an accusation of misconduct, and a small trial is not automatically worthless. The practical implication is about confidence: early studies can motivate further research without supporting a strong claim that the outcome will generalize. A review's cautious conclusion should remain visible next to favorable findings, not confined to a final disclaimer.
It also matters that the review considered glycine across many physiological systems and populations. Findings about another condition, another administration route or a much longer exposure should not be imported into a sleep-product argument. The appropriate evidence for this article is the sleep and next-day research itself, with its own limits. [4]
5. Japanese product labels: examples, not trial equivalents
These three stocked products illustrate why the name “glycine” is not enough to compare formulas. The amounts below were checked against the manufacturers' current information and matched to the store products by JAN barcode. All amounts refer to the complete serving shown, not one tablet.[5][6][7]
The combination and amounts differ from the trial regimen. The product name does not establish that this formula produces the studied outcomes.
The 5.1 g figure for the FINE powder is the weight of the whole stick's contents; 4,000 mg is its listed glycine amount. These are different quantities. Similarly, the tablet products' amounts apply to the stated multi-tablet servings.
The cited trials did not test these exact retail formulations. Do not increase tablet counts to imitate a trial dose: that would also increase the other ingredients and could exceed the product's directions. Check the current package for ingredients, allergens, instructions, and cautions.
Why the powder-versus-tablet comparison needs more than a number
The FINE powder has a glycine-centered name and a larger listed amount, but it also contains four other listed components. The FINE tablet formula has a three-to-six-tablet serving range, so its ingredient amounts legitimately appear as ranges. The Noguchi product lists a fixed two-tablet serving. Those label differences are useful for understanding composition and routine—not for predicting which product will produce the strongest sleep result.
Do not assume that combining glycine with GABA recreates or strengthens the glycine-only evidence. The original trials were not head-to-head comparisons of these blends, and the additional ingredients introduce another factual boundary. Similarly, a powder or tablet form does not establish a faster onset or superior absorption from the information presented here.
Shipping from Japan and having Japanese branding also do not establish that every raw ingredient originated in Japan. For example, Noguchi's current manufacturer information identifies its GABA ingredient as manufactured in China. Origin, serving composition and clinical evidence are separate questions; verify the particular fact you need rather than inferring it from the brand name. [7]
6. Questions to ask before drawing a conclusion
Do these studies show that glycine increases deep sleep?
Not as a general conclusion. The 2007 PSG study found a shorter latency to slow-wave sleep but no change in the proportions of the different sleep stages. Earlier onset is not the same as a larger amount. [2]
Does the 2012 study show less fatigue every day?
No. The fatigue visual analog scale favored glycine on day one, but the day-three subjective results were not significant. Sleepiness and fatigue were separate outcomes, and the analysis included only seven men under temporary sleep restriction. [3]
Is a higher glycine label amount necessarily better?
No. A label states composition, not comparative effectiveness. The trials discussed here do not establish an optimal amount for an individual or superiority of the listed retail products.
How should I use this evidence when comparing products?
Keep three questions separate: what the study measured, what the product contains, and whether its directions and cautions suit your situation. The research regimen is not a substitute for the product label or professional advice.
Why can the sleepiness bars be lower without a significant result?
The statistical test considers variability and the study design, not only the distance between means. With seven participants, the observed pattern may be too uncertain to establish the outcome. This is not the same as proving there is no possible effect. The original day-one treatment result was p = 0.098. [3]
Can a body-temperature mechanism prove that a supplement works?
No. The Japanese-linked mechanism study discussed here involved rats and experimental manipulations. It offers a hypothesis to investigate, not confirmation of a human result from a retail serving. Human clinical outcomes and product-specific evidence are still separate requirements. [8]
Bottom line: Glycine's early Japanese sleep research is more nuanced than a simple “sleep better” promise. Some subjective ratings, sleep-onset timings, and next-day measures were favorable; other outcomes were unchanged or inconclusive. Use the studies to understand the questions being tested, then compare the powder label and the two tablet labels above without treating them as clinically interchangeable.
This article is educational, not personal medical advice. Persistent sleep difficulties or daytime impairment deserve professional assessment. Follow the current product directions and seek advice before use if a label caution applies to you.
Sources and label references
Inagawa K, et al. “Subjective effects of glycine ingestion before bedtime on sleep quality.” Sleep and Biological Rhythms. 2006;4:75–77. Original study.
Yamadera W, et al. “Glycine ingestion improves subjective sleep quality in human volunteers, correlating with polysomnographic changes.” Sleep and Biological Rhythms. 2007;5:126–131. Original study.
Bannai M, et al. “The effects of glycine on subjective daytime performance in partially sleep-restricted healthy volunteers.” Frontiers in Neurology. 2012;3:61. Open-access study. Human results are distinguished here from the paper's separate animal experiments.
Soh et al. “The effect of glycine administration on the characteristics of physiological systems in human adults: A systematic review.” GeroScience. 2024;46:219–239; published online October 18, 2023. Systematic review.
Noguchi Medical Research Institute. GABA & Glycine, JAN 4562193142753. Manufacturer serving and composition information.
Kawai N, et al. “The Sleep-Promoting and Hypothermic Effects of Glycine are Mediated by NMDA Receptors in the Suprachiasmatic Nucleus.” Neuropsychopharmacology. 2015;40:1405–1416; published online December 23, 2014. Original animal-mechanism study. This is not a human efficacy trial.
Product-label information checked October 6, 2026. Formulas and packaging may change; use the current package for your final checks.
