Natto: modest trials, one clear warning
Natto is the Japanese breakfast food with a drug warning attached. Start with what has actually been randomized. In 36 non-diabetic adults, four out of five of them men, natto rich in gamma-polyglutamic acid held the first half hour of blood glucose after a test meal to 351.2 against 505.3 mg per minute per dL for natto low in it. That came from a single meal loading rather than a stretch of ordinary eating. In 11 healthy adults a breakfast of white rice with 1.8 oz (50 g) of natto, yam and okra held peak glucose at 6.0 against 123 mg/dL (6.8 mmol per liter), and that belongs to the whole sticky meal rather than to the beans. Everything else here is observational, or a single case. The warning is the part to read first: a systematic review of warfarin users set an acceptable vitamin K intake of 25 to 325 micrograms a day and named fermented soybean, and any food made with it, as the only ordinary food to avoid entirely. That range is what a review of published practice found acceptable, not a limit set by a drug regulator.
Every statement here is tied to a source. The badge says what kind of evidence stands behind it: A is a meta-analysis or systematic review, E is the position of an expert body without its own analysis. How we verify →
The numbers, per 3.5 oz (100 g)
| USDA entry | kcal | Protein | Fat | Carbs | Fiber |
|---|---|---|---|---|---|
| Natto | 211 | 19.4 g | 11 g | 12.7 g | 5.4 g |
These are the USDA entries that are natto, not foods made from it; they differ in cut, part, pack or preparation. Linked names have a page with the full profile and per-portion figures.
As people eat it, per 3.5 oz (100 g)
| As eaten | kcal | Protein | Fat | Carbs | Fiber |
|---|---|---|---|---|---|
| Natto | 211 | 19.4 g | 11 g | 12.7 g | 5.4 g |
USDA Food and Nutrient Database for Dietary Studies (FNDDS, release 2024-10-31): natto as it is prepared and served, from the recipes behind the national dietary survey, rather than a single laboratory sample.
How much is actually in it
- USDA lists natto at 8.6 mg iron, 0.667 mg copper and 19.4 g protein per 3.5 oz (100 g). — USDA reference entry for natto, per 3.5 oz (100 g) (n = 1 record) Evidence E sourceIron 8.6 mg, copper 0.667 mg, protein 19.4 g per 3.5 oz (100 g)
- No sampling variability is published for natto: USDA holds one legacy entry and no Foundation Foods samples, and the UK composition table has no natto entry at all. — USDA Foundation Foods (395 foods) and UK CoFID (2886 foods), searched for natto (n = 0 samples) Evidence E sourceNo range, no second independent value
What a real portion looks like
- In the US food coding dataset one cup of natto weighs 6.2 oz (176 g). — US dietary survey food coding, WWEIA category Soy-based condiments (n = 1 food code) Evidence E source1 cup = 176 g, food code 41420200
What your body absorbs
- Natto-derived menaquinone-7 and synthetic vitamin K1 both peak in serum 4 hours after intake, but MK-7 accumulates to 7 to 8 fold higher serum levels during prolonged intake. — healthy volunteers, supplement comparison of K1 and natto-derived MK-7 (n = not stated in abstract) Evidence C source7 to 8 fold higher serum MK-7 during prolonged intake, peak at 4 hours for both forms
- By the NIH fact sheet, data on the bioavailability of different forms of vitamin K from food are very limited, and long-chain menaquinones may be absorbed better than phylloquinone from green vegetables. — general adult population, agency review of absorption data Evidence E sourceNo quantified absorption figure for menaquinones from food
What cooking and storage change
- Boiling natto cuts its menaquinone-7 content from about 660 to about 79 ng/mL and blunts the plasma MK-7 rise after eating it. — adult volunteers eating untreated or boiled natto, plasma vitamin K measured at 5, 8, 24 and 48 hours (n = not stated in abstract) Evidence C sourceMK-7 content 660.40 plus minus 65.32 to 78.50 plus minus 11.12 ng/mL; plasma MK-7 at 5 hours 14.54 ng/mL untreated vs 4.02 ng/mL boiled
- The USDA retention factor table has no cooking method for natto and does not cover the food at all, so nutrient losses from cooking natto have no dataset value. — USDA Table of Nutrient Retention Factors, Release 6, searched for natto (n = 0 methods) Evidence E sourceNo retention factor available
What it does to your health
- A meta-analysis of 3 prospective studies in 1658 perimenopausal women links natto intake with higher femoral neck bone mineral density (WMD 0.42, 95% CI 0.05 to 0.79) and with no change at the hip joint. — perimenopausal women, 3 prospective studies, 1658 non-overlapping participants (n = 1658) Evidence C sourceFemoral neck WMD 0.42, 95% CI 0.05 to 0.79, I2 72%; hip joint no significant difference
- A 2025 meta-analysis of 6 observational and quasi-experimental studies in 2327 Japanese adults links habitual natto intake with much higher serum MK-7 (d 2.10, 95% CI 1.55 to 2.66) and with modestly greater bone mineral density (d 0.35, 95% CI 0.21 to 0.48). — Japanese adults, habitual natto consumers vs low or non-consumers, 6 studies (n = 2327) Evidence C sourceSerum MK-7 d 2.10, 95% CI 1.55 to 2.66; BMD d 0.65, 95% CI 0.09 to 1.21, sensitivity analysis d 0.35, 95% CI 0.21 to 0.48
- In a randomized crossover trial in 36 adults, natto rich in gamma-polyglutamic acid lowered the 0 to 30 minute blood glucose IAUC to 351.2 mg per min per dL against 505.3 after low gamma-polyglutamic acid natto. — 36 non-diabetic adults, men aged 20 to 70 and postmenopausal women under 70, BMI 18.5 to 24.9, single meal loading (n = 36) Evidence B sourceBlood glucose IAUC 0 to 30 min 351.2 plus minus 28.0 vs 505.3 plus minus 36.0 mg per min per dL, p below 0.001
- In a randomized crossover trial in 11 healthy adults, a white rice breakfast with 50 g natto and viscous vegetables cut peak glucose to 108 mg/dL (6.0 mmol/L) against 6.8 after a matched non-viscous control meal. — 11 healthy adults, single test meal of 200 g boiled white rice with 50 g natto, 60 g yam and 40 g okra (n = 11) Evidence B sourcePeak glucose 108 vs 123 mg/dL (6.0 vs 6.8 mmol/L) and peak insulin 262 vs 360 pmol/L; glucose and insulin IAUC over 0 to 120 min reduced by 28% and 27%
- Observational data link eating natto several packs a week with lower all-cause mortality in 1548 Japanese men aged 65 and over followed a mean 12.0 years (HR 0.603, 95% CI 0.441 to 0.825). — 1548 community-dwelling Japanese men aged 65 and over, mean follow-up 12.0 years, 430 deaths (n = 1548) Evidence C sourceHR 0.603, 95% CI 0.441 to 0.825 for several packs per week; HR 0.786, 95% CI 0.539 to 1.145 for one pack a day or more
Safety, limits and interactions
- A systematic review of warfarin users put the acceptable vitamin K intake at 25 to 325 micrograms a day and named fermented soybean, and foods containing it, the only ordinary foods to be avoided entirely. — warfarin users, 16 studies on the upper limit and 6 on amounts below it, from 1310 citations (n = 22 studies) Evidence C sourceAcceptable range 25 to 325 micrograms a day, maximum daily variation 292 micrograms, optimum about 150 micrograms a day
- By the NIH fact sheet, vitamin K has a serious and potentially dangerous interaction with warfarin and similar anticoagulants, and people taking them must keep vitamin K intake consistent. — people taking warfarin, phenprocoumon, acenocoumarol or tioclomarol Evidence E sourceNo numeric threshold given by the agency
- A case report identifies poly gamma-glutamic acid, the polymer of natto mucilage, as the allergen behind late onset anaphylaxis that began about 3 hours after eating. — one man aged 38 with repeated late onset anaphylaxis after natto, skin prick tested (n = 1) Evidence D sourcePositive skin prick test to natto and to poly gamma-glutamic acid; reaction 3 hours after ingestion
What people believe that the data does not support
- The EU register lists the heart health claim for Bacillus subtilis Natto as non-authorized, so it cannot be used on labels in the EU. — EU labeling of foods carrying Bacillus subtilis Natto (n = 1 claim record) Evidence E sourceStatus Non-authorized
- The EU register also lists the gut health claims for Bacillus subtilis Natto as non-authorized, and none of the three claims filed for this substance was authorized. — EU labeling of foods carrying Bacillus subtilis Natto (n = 3 claim records, 0 authorized) Evidence E sourceStatus Non-authorized for all three
Who this works differently for
- Observational data link the highest tertile of natto intake with lower atrial fibrillation risk in women (HR 0.44, 95% CI 0.24 to 0.80) and with no change in men, among 5278 adults followed a mean 12.6 years. — 5278 Japanese males and females aged 30 to 90, 66487 person-years, 222 incident atrial fibrillation cases (n = 5278) Evidence C sourceFemales highest vs lowest tertile HR 0.44, 95% CI 0.24 to 0.80; males HR 0.97, 95% CI 0.65 to 1.43
- Case reports describe Bacillus subtilis var. natto bloodstream infection in three habitual natto eaters with impaired gut barrier or immune status, one of them at 36 weeks of pregnancy. — 3 patients: a man aged 63 on maintenance hemodialysis, a woman aged 35 at 36 weeks of gestation, a man aged 62 with colorectal cancer (n = 3) Evidence D sourceBacteremia resolved on targeted antimicrobial therapy in all three; the pregnant patient underwent emergency cesarean delivery
- Observational data link natto intake with higher odds of Bacillus subtilis bacteremia in children receiving chemotherapy for cancer (odds ratio 3.3, 95% CI 1.1 to 9.6). — 23 children under 15 with Bacillus subtilis bacteremia during chemotherapy, matched to 92 controls, Tokyo 2012 to 2020 (n = 115) Evidence C sourceOdds ratio 3.3, 95% CI 1.1 to 9.6 after controlling for neutropenia; 78% of cases and 50% of controls ate natto within 7 days
- Observational data from 650 Japanese mother and baby pairs link daily maternal natto intake during pregnancy with lower incidence of infant eczema at 6 months. — 650 mother and baby pairs, birth cohort at two hospitals in Chiba, eczema assessed at 6 months (n = 650) Evidence C sourceLower eczema incidence with everyday intake vs 2 to 3 times a week or less, p = 0.020
- Cross-sectional data on 616 Japanese adults aged 65 and over link natto eating habits with lower odds of mild cognitive impairment (odds ratio 0.63, p = 0.025) and of depressive symptoms (odds ratio 0.63, p = 0.037). — 616 Japanese adults aged 65 and over, adjusted for age, sex, education, frailty, dietary diversity, tofu and miso habits (n = 616) Evidence C sourceMild cognitive impairment odds ratio 0.63, p = 0.025; depressive symptoms odds ratio 0.63, p = 0.037
What is still unknown
- No interventional trial of natto is registered on ClinicalTrials.gov, FDA has authorized no health claim naming natto in 21 CFR 101, and no drug interaction record is indexed for natto or for vitamin K in the interaction layer. — registry searches for natto on 2026-09-25 (n = 0 records in each registry) Evidence E sourceNo registered trial, no US authorized claim, no indexed interaction record
The bottom line
USDA holds one record for natto and no sampled range, and the UK composition table has no entry at all, so 8.6 mg of iron, 0.667 mg of copper and 19.4 g of protein per 3.5 oz (100 g) stand on a single line of data. A cup is counted as 176 g in the US survey database. The vitamin K in it is menaquinone-7. In a comparison in healthy volunteers it peaked in serum 4 hours after intake, the same as synthetic K1, and built up to 7 to 8 times higher levels on continued intake, in a report that does not say how many volunteers took part. Boiling takes most of it away, from about 660 to about 79 ng per mL of MK-7, with the plasma rise at 5 hours falling from 14.54 to 4.02 ng per mL, in a small uncontrolled test whose size the abstract also leaves out. Observational data link natto with bone density. Three prospective studies in 1658 perimenopausal women gave a weighted mean difference of 0.42 at the femoral neck, interval 0.05 to 0.79, with heavy disagreement between the studies and nothing at the hip. A 2025 pooling of 6 studies in 2327 Japanese adults put the bone effect at 0.35 standardized units, interval 0.21 to 0.48, and its authors rate that part of their own evidence low. In 1548 Japanese men aged 65 and over, several packs a week went with a hazard ratio of 0.603 for death from any cause, while a pack a day or more did not, at 0.786 with an interval from 0.539 to 1.145. Other single observational studies in Japanese populations link natto with lower atrial fibrillation in women, hazard ratio 0.44, and with no change in men, and with lower odds of mild cognitive impairment in people over 65, odds ratio 0.63. Now the cautions, which are specific. If you take warfarin or a similar anticoagulant, this is the food that review singled out to skip, together with anything made with it, and NIH calls that interaction serious and potentially dangerous. Three people with a damaged gut barrier or weakened immunity, one of them a woman of 35 at 36 weeks of pregnancy, developed bloodstream infection with the natto bacterium itself. Among children on chemotherapy at one Tokyo hospital, 23 cases against 92 controls, eating natto in the previous week carried 3.3 times the odds of Bacillus subtilis bacteremia, which is odds rather than risk. One man of 38 had anaphylaxis about 3 hours after eating, caused by the mucilage polymer rather than by the soybean, and that polymer turns up in other foods too. Nothing here measures how often people react to the soy in natto itself. Against all that, the EU authorized none of the three claims filed for Bacillus subtilis Natto, no interventional trial of natto is registered on ClinicalTrials.gov, and the United States has authorized no claim naming it.
More from the same food group (legumes and legume products)
- Papad: a salty wafer that is small enough to count
- Peanut butter: the number on the label came from almonds
- Peanut flour: a pantry protein that allergy clinics dose by the milligram
- Miso: a heavy salt carrier with mixed evidence on blood pressure and cancer
- Meatball: the side dish moved the meal more than the meatballs did
- Lupins: modest cholesterol results and two real cautions
Sources
- natto-r1-01 · food composition dataset
Natto, FDC 172443, sr_legacy, Legumes and Legume Products: copper 0.667 mg, iron 8.6 mg, protein 19.4 g per 100 g
USDA FoodData Central, SR Legacy, FDC 172443, 2019 · checked 2026-09-25 - natto-r1-02 · dataset search
variability_search natto: no Foundation Foods sample (Foundation covers 395 foods, SR Legacy carries no variability); cofid_search natto: no entry among 2886 UK foods
USDA FoodData Central Foundation Foods and UK CoFID, searched 2026-09-25 · checked 2026-09-25 - natto-r1-07 · portion dataset
Natto, WWEIA category Soy-based condiments, FNDDS code 41420200, fdc_id 2707440: 1 cup, measure 10205, 176 g
USDA FNDDS, food code 41420200, FDC 2707440 · checked 2026-09-25 - natto-r1-03 · comparative clinical trial
Both K(1) and MK-7 were absorbed well, with peak serum concentrations at 4 hours after intake. A major difference between the 2 vitamin K species is the very long half-life time of MK-7, resulting in much more stable serum levels, and accumulation of MK-7 to higher levels (7- to 8-fold) during prolonged intake.
Blood, 2007 · checked 2026-09-25 - natto-r1-04 · fact sheet
Data on the bioavailability of different forms of vitamin K from food are very limited [1]. ... Limited research suggests that long-chain MKs may have higher absorption rates than phylloquinone from green vegetables [7].
NIH Office of Dietary Supplements, Vitamin K Fact Sheet for Health Professionals · checked 2026-09-25 - natto-r1-05 · clinical trial of cooking methods
Boiling significantly reduced the Bacillus subtilis count to 5.1+/-0.3 Log cfu/mL, and concomitantly reduced the content of menaquinone-7 (MK-7), which is a form of vitamin K synthesized by Bacillus subtilis, from 660.40+/-65.32 ng/mL to 78.50+/- 11.12 ng/mL.
J Nutr Sci Vitaminol (Tokyo), 2006 · checked 2026-09-25 - natto-r1-06 · dataset search
USDA Table of Nutrient Retention Factors, Release 6 (2007), natto: no method at all, and FoodData Central does not know this food either
USDA Table of Nutrient Retention Factors, Release 6, 2007 · checked 2026-09-25 - natto-r1-08 · meta-analysis of prospective studies
In addition, natto significantly improved the BMD of the femoral neck in a cohort study and randomized control study (P=0.03, WMD=0.42). 95% CI:0.05-0.79, I2= 72%); (P < 0.0001, WMD=0.16; 95% CI:0.08-0.24), respectively. However, all studies demonstrated that natto has no improvement effects on a hip joint (BMD).
Curr Med Chem, 2021 · checked 2026-09-25 - natto-r1-09 · meta-analysis of observational studies
Natto intake is associated with significantly elevated serum MK-7 [d = 2.10, 95% CI (1.55, 2.66) after sensitivity analysis removing an outlier], increased OC [d = 0.26, 95% CI (0.08, 0.43)], decreased ucOC [d = -0.50, 95% CI (-0.74, -0.26)], and modestly greater BMD across sites [d = 0.65, 95% CI (0.09, 1.21); sensitivity analysis: d = 0.35, 95% CI (0.21, 0.48)].
Front Nutr, 2025 · checked 2026-09-25 - natto-r1-10 · randomized crossover trial
The levels of blood glucose IAUC at 0 to 30 min, the primary outcome of the present study, of the HPGA meal (351.2 ± 28.0 (est. mean ± SE) mg·min/dL) were significantly lower than those generated by the LPGA meal (505.3 ± 36.0 mg·min/dL) (p < 0.001).
Nutrients, 2020 · checked 2026-09-25 - natto-r1-11 · randomized crossover trial
Peak glucose and insulin concentrations after the test meal (6.0 mmol/L and 262 pmol/L) were significantly lower than after the control meal (6.8 mmol/L and 360 pmol/L).
Asia Pac J Clin Nutr, 2008 · checked 2026-09-25 - natto-r1-12 · prospective cohort study
Compared with 'no consumption' of natto, HRs for 'several packs/week' and 'pack/day or more' were 0.603 (95 % confidence interval (CI) 0.441 to 0.825) and 0.786 (95 % CI 0.539 to 1.145), respectively.
Clin Nutr ESPEN, 2025 · checked 2026-09-25 - natto-r1-14 · systematic review
The intake of vitamin K in warfarin patients was acceptable in the range of 25-325 μg/day, with a maximum daily variation of 292 μg, and a value of 150 μg/day seemed optimum. When these results were applied to usual foods, except for dietary supplements or health foods, the only prohibited foods were fermented soybean (natto) and foods containing it, while green leafy vegetables could be acceptable if their intake is limited.
Shokuhin Eiseigaku Zasshi, 2015 · checked 2026-09-25 - natto-r1-15 · fact sheet
Vitamin K can have a serious and potentially dangerous interaction with anticoagulants such as warfarin (Coumadin) as well as phenprocoumon, acenocoumarol, and tioclomarol, which are commonly used in some European countries [7,8]. These drugs antagonize the activity of vitamin K, leading to the depletion of vitamin K-dependent clotting factors. People taking warfarin and similar anticoagulants need to maintain a consistent intake of vitamin K from food and supplements because sudden changes in vitamin K intakes can increase or decrease the anticoagulant effect [45].
NIH Office of Dietary Supplements, Vitamin K Fact Sheet for Health Professionals · checked 2026-09-25 - natto-r1-18 · case report
Therefore, he was diagnosed with late-onset anaphylaxis to PGA contained in natto and the cold Chinese noodle soup.
Allergol Int, 2011 · checked 2026-09-25 - natto-r1-19 · regulatory register
POL-HC-6671 Bacillus subtilis Natto • Helps support a healthy heart • Helps maintain a healthy heart Non-authorised
EU Register on nutrition and health claims, claim POL-HC-6671, snapshot 2026-09-21 · checked 2026-09-25 - natto-r1-20 · regulatory register
POL-HC-6673 Bacillus subtilis Natto • For good intestinal health • Helps keep your digestive system in balance • Helps gastrointestinal tract comfort • Maintains a healthy intestinal environment and functionality • Maintain the balance of microflora • Promotes and maintains intestinal hea Non-authorised
EU Register on nutrition and health claims, claim POL-HC-6673, snapshot 2026-09-21 · checked 2026-09-25 - natto-r1-13 · prospective cohort study
Compared with the lowest tertile of natto intake group (T1), we observed a decreased risk of AF in the highest tertile of the natto intake group (T3) in females (HR: 0.44; 95% CI: 0.24, 0.80). In contrast, there was no association between natto intake and AF risk in males (T3 compared with T1: HR: 0.97; 95% CI: 0.65, 1.43).
J Nutr, 2026 · checked 2026-09-25 - natto-r1-16 · case series
We report three cases of B. subtilis var. natto bacteremia confirmed by biotin operon analysis: a 63-year-old man receiving maintenance hemodialysis, a 35-year-old woman at 36 weeks of gestation, and a 62-year-old man whose bacteremia led to the diagnosis of colorectal cancer.
IDCases, 2026 · checked 2026-09-25 - natto-r1-17 · case-control study
Multivariable logistic regression tests after controlling for neutropenia revealed that B. subtilis bacteremia was associated significantly with natto intake at odds ratio 3.3 (95% CI: 1.1-9.6).
J Infect Chemother, 2023 · checked 2026-09-25 - natto-r1-21 · birth cohort study
For natto, incidence of babies' eczema was significantly lower in the group with everyday intake than those eating it 2-3 times a week or less (p = 0.020).
Allergol Int, 2014 · checked 2026-09-25 - natto-r1-22 · cross-sectional study
The logistic regression analysis showed that natto consumption habits were negatively associated with MCI, even after adjustment (odds ratio [OR]: 0.63; p = .025). Natto consumption habits were also negatively associated with depressive symptoms (OR: 0.63; p = .037).
Nutr Res, 2026 · checked 2026-09-25 - natto-r1-23 · registry search
ctgov_search natto: nothing; claims_search natto and vitamin K in 21 CFR 101: nothing; interact_search natto and vitamin K: nothing
ClinicalTrials.gov and 21 CFR part 101, searched 2026-09-25 · checked 2026-09-25
Review. Reviewed on 2026-09-25 for evidence level, dose and population context, and claims a reader could misread. Bioma Learn has no human medical reviewer at this time, and we say so rather than invent one. Methodology. This is information, not medical advice.