Soybeans: the hormone scare forty trials could not find

The question people actually ask about soybeans is whether they act like estrogen. Forty randomised trials on 3285 postmenopausal women, using a median 75 mg of isoflavones a day for a median of 24 weeks, found no change in any measure of estrogenicity. Endometrial thickness differed by 0.22 mm at most, with a confidence interval from -0.45 to 0.01 mm, and the review authors rated that evidence high to moderate. The 75 mg is a median reported isoflavone dose, so the trials behind it used both more and less, and the review does not say how high the highest dose went. In men the answer comes out the same. Fifteen placebo-controlled treatment groups showed no change in total testosterone, sex hormone-binding globulin or free testosterone from soy foods or isoflavone supplements, though that analysis was published in 2010 and measured hormone concentrations rather than fertility. What soy protein does do is smaller and more specific than the reputation on either side. In 18 trials on 522 people with chronic kidney disease, replacing animal protein with soy protein lowered total cholesterol by about 21 mg/dL and LDL by about 8 mg/dL. The authors call that estimate imprecise, because the trials and the numbers of people in them were few. That number belongs to people with chronic kidney disease who replaced animal protein with soy, not to a healthy adult adding tofu to dinner, and the only cholesterol figure on this page for adults in general is the 5.0 mg/dL from fermented soy products.

The numbers, per 100 g

What your body absorbs

• In 12 trials on 261 young adults doing resistance training, neither soy nor whey protein changed lean body mass, although both raised peak plasma essential amino acids. — trained and untrained adults aged 18 to 30 years doing resistance exercise training, 12 studies (n = 261) Evidence A source
  No significant effect on lean body mass for soy or whey; whey bench press mean difference 8.87 (95% CI 5.95 to 11.79) and squat 9.60 (95% CI 5.61 to 13.60), soy not significant on either
• Measured by indicator amino acid oxidation across 10 studies of legumes including soy protein, lysine availability ran above 80 per cent while methionine stayed below 80 per cent. — 10 studies, 7 in humans (school-aged children and adult men) and 3 in animals, legumes including soy protein, peas, faba beans, chickpeas, lentils and black beans (n = 10 studies) Evidence C source
  Metabolic availability of lysine generally above 80 per cent, methionine below 80 per cent
• Across 59 studies reviewed, fermenting soybeans cut phytate and tannin content and raised protein digestibility and mineral availability. — 59 peer-reviewed studies on finger millet and soybean fermentation, published 2000 to 2025 (n = 59 studies) Evidence D source
  Direction only: antinutritional factors down, protein digestibility and mineral bioavailability up; no pooled size reported

What cooking and storage change

• Across 15 randomised trials, fermented soy products lowered fasting plasma glucose by 6.4 mg/dL and total cholesterol by 5.0 mg/dL, while body mass index moved by only 0.14 units. — adults in 15 randomised controlled trials, screened from 2205 records (n = 15 trials) Evidence A source
  Fasting plasma glucose -6.39 mg/dL (95% CI -10.38 to -2.40); total cholesterol -5.0 mg/dL (95% CI -6.60 to -3.39); BMI -0.14 kg/m2 (95% CI -0.28 to -0.01); waist circumference -1.50 cm (95% CI -2.94 to -0.07); visceral fat -692.17 mm2 (95% CI -1011.58 to -372.77)

Safety, limits and interactions

• The FDA-approved label for levothyroxine sodium names soybean flour among the foods that can bind the drug and cut its absorption from the gut. — people taking levothyroxine sodium, per the approved product label effective 2026-03-24 Evidence E source
  Direction only: reduced gastrointestinal absorption of levothyroxine sodium; the label gives no size and no dose of soy at which it happens
• Untreated natto pushed plasma menaquinone-7 from 1.86 to 14.54 ng/mL five hours after eating, while boiling the same fermented soybeans held the rise to 4.02 ng/mL. — adult volunteers eating natto or boiled natto, plasma vitamin K measured at 5, 8, 24 and 48 hours (n = volunteers, number not stated in the abstract) Evidence C source
  Untreated natto: plasma MK-7 1.86 plus or minus 1.51 to 14.54 plus or minus 4.12 ng/mL at 5 h, still 5.37 ng/mL at 48 h; boiled natto: 1.61 to 4.02 ng/mL at 5 h; boiling cut natto MK-7 content from 660.40 to 78.50 ng/mL

What people believe that the data does not support

• In 40 randomised trials on 3285 postmenopausal women, soy isoflavones at a median 75 mg per day over a median of 24 weeks changed no measure of estrogenicity, with endometrial thickness differing by 0.22 mm at most. — postmenopausal women, 40 trials, 52 trial comparisons, median 24 weeks (n = 3285) Evidence A source
  Endometrial thickness mean difference -0.22 mm (95% CI -0.45 to 0.01); vaginal maturation index 2.31 (-2.14 to 6.75); FSH -0.02 IU/L (-2.39 to 2.35); estradiol 1.61 pmol/L (-1.17 to 4.38)
• Pooling 15 placebo-controlled treatment groups in men, soy foods and isoflavone supplements left testosterone, sex hormone-binding globulin and free testosterone unchanged. — adult men, 15 placebo-controlled treatment groups with baseline and ending measures, plus 36 treatment groups in simpler models (n = 15 treatment groups) Evidence A source
  No significant change in total testosterone, SHBG, free testosterone or free androgen index in any statistical model
• The European Union register holds 19 claims about soy protein and soy isoflavones and has authorised none of them, including the cholesterol claim that the United States still permits. — EU Register on nutrition and health claims, snapshot 2026-09-21: 7 claims under soy protein and 12 under soy isoflavones, all non-authorised (n = 19 claims) Evidence E source
  0 authorised, 19 rejected under these two substance names

Who this works differently for

• In 18 trials on 522 people with chronic kidney disease, replacing animal protein with soy protein lowered total cholesterol by about 21 mg/dL and LDL by about 8 mg/dL. — people with chronic kidney disease, 18 randomised trials (n = 522) Evidence A source
  Total cholesterol -20.55 mg/dL (95% CI -38.25 to -2.85); LDL -8.26 mg/dL (95% CI -13.35 to -3.17); proteinuria -140.53 mg/day (95% CI -205.83 to -75.23); no change in creatinine, triglycerides, calcium, CRP, HDL, uric acid or phosphorus
• In a network meta-analysis of 60 trials on 6284 healthy women, soybean protein raised total body bone mineral density by 0.03 g per square centimetre, while whole soy foods beat no treatment on nothing. — healthy women, mean age 54.2 years, 60 randomised controlled trials published 1995 to 2022, interventions averaging 11.3 months (n = 6284) Evidence A source
  Soybean protein total body BMD +0.03 g/cm2 (95% CI 0.01 to 0.05); casein plus whey +0.04 g/cm2 (95% CI 0.01 to 0.06); at food level no dairy or soy intervention beat control

What is still unknown

• The United States health claim linking soy protein to lower heart disease risk has stood since 1999, and in 2017 the FDA proposed to revoke it, a proposal still unresolved in 2026. — United States food labelling, 21 CFR 101.82, proposed rule of 31 October 2017 Evidence E source
  Regulatory status: the claim remains in force in 21 CFR 101.82 while the proposal to revoke it is unresolved
• An umbrella review of 23 meta-analyses of observational research and 1 of trials found soy foods mostly beneficial, and left fermented soy products unresolved, with some studies pointing to harm. — 23 meta-analyses of observational research covering 39 unique health outcomes, plus 1 meta-analysis of randomised trials covering 4 outcomes, searched to April 2024 (n = 24 meta-analyses) Evidence C source
  Direction only: benefit concentrated in cancer and cardiometabolic outcomes; fermented soy inconclusive, with some studies suggesting adverse effects

The bottom line

Fermented soy carries the clearest numbers here and the least clear meaning. Across 15 randomised trials it lowered fasting plasma glucose by 6.4 mg/dL, with a confidence interval from -10.4 to -2.4, and total cholesterol by 5.0 mg/dL, while body mass index moved 0.14 units. That review pooled several different fermented foods and supplements, so the figure belongs to a category and not to any one food. An umbrella review of 24 meta-analyses, almost all of them observational, called soy foods mostly beneficial and left fermented soy unresolved, with some of the underlying studies pointing towards harm. Fermentation as chemistry is better described than fermentation as nutrition. A narrative review of 59 laboratory and compositional studies reports that it cuts phytate and tannin and raises protein digestibility and mineral availability, giving the direction without any magnitude and without a human absorption endpoint. On protein quality, a scoping review of 10 studies of legumes as a class, 3 of them in animals, put lysine availability above 80 per cent and methionine below it. If you are eating soy for muscle, 12 trials on 261 adults aged 18 to 30 in resistance training found no change in lean body mass from soy or from whey, and the strength gains in that analysis belonged to whey alone. For bone, a network meta-analysis of 60 trials on 6284 healthy women averaging 54 years found that soybean protein raised total body bone mineral density by 0.03 g/cm2, while at the level of whole foods no soy intervention beat the control on anything. Regulators read all of this differently from each other. The European Union has 19 claims filed under soy protein and soy isoflavones and has authorised none of them, which records that the assessors were unconvinced at the time rather than that the effects are absent. The United States still allows the soy protein and heart disease claim in 21 CFR 101.82, and the FDA proposal to revoke its own claim, published in 2017, was still unresolved in 2026. The safety axis on this page holds two narrow things. The FDA-approved label for levothyroxine sodium lists soybean flour among foods that may bind the drug and reduce its absorption from the gut. That label names flour and infant soy formula rather than whole cooked soybeans, and it gives no dose and no size of effect. Natto is also a vitamin K food. In one feeding test with no randomisation and no control arm, whose abstract never says how many volunteers took part, untreated natto raised plasma menaquinone-7 from 1.86 to 14.54 ng/mL five hours after eating. Boiling the same natto held that rise to 4.02 ng/mL and cut its MK-7 content from 660 to 79 ng/mL. Those volunteers were not warfarin patients, so if you take warfarin, natto is a question for the person managing your dose. Nothing on this page speaks to soy allergy, to children or to pregnancy, and that silence is not reassurance.

Sources

1. sybn-r4-06 · systematic review and meta-analysis of RCTs
   A total of 12 studies with 261 participants were included after full-text screening with the main reason for exclusion being wrong population. The analysis found no significant effect of either whey or soy protein supplementation on LBM.
   J Diet Suppl, 2026 · checked 2026-09-24
2. sybn-r4-07 · scoping review
   Of the ten studies included, three were conducted on animals, while the remaining seven involved human participants, specifically school-aged children and adult men...The MA of lysine in legumes was generally high (>80%), whereas methionine showed relatively lower availability (<80%).
   Crit Rev Food Sci Nutr, 2026 · checked 2026-09-24
3. sybn-r4-08 · narrative systematic review
   Fermentation reduces antinutritional factors such as phytates and tannins, improves protein digestibility, and increases mineral bioavailability.
   Front Nutr, 2026 · checked 2026-09-24
4. sybn-r4-04 · systematic review and meta-analysis of clinical trials
   A total of 2205 records were found, of which 15 RCTs were eligible...Overall, fermented soy may confer health benefits on certain metabolic outcomes, anthropometric indices, and body composition.
   Food Funct, 2025 · checked 2026-09-24
5. sybn-r4-09 · regulatory product label
   Soybean flour, cottonseed meal, walnuts, and dietary fiber may bind and decrease the absorption of levothyroxine sodium from the gastrointestinal tract.
   Levothyroxine sodium, FDA-approved label, drug interactions, effective 2026-03-24 · checked 2026-09-24
6. sybn-r4-10 · clinical feeding study
   In conclusion, boiled natto did not cause a marked increase in the plasma concentration of vitamin K in subjects who consumed it. Thus, patients on warfarin may be able to eat boiled natto without ill effects.
   J Nutr Sci Vitaminol (Tokyo), 2006 · checked 2026-09-24
7. sybn-r4-01 · systematic review and meta-analysis of RCTs
   We included 40 trials (52 trial comparisons, n = 3285) assessing the effect of a median reported dose of 75 mg/d of soy isoflavones in substitution for non-isoflavone controls over a median of 24 wk...Soy isoflavones had no statistically significant effect on any measure of estrogenicity
   Adv Nutr, 2025 · checked 2026-09-24
8. sybn-r4-02 · meta-analysis
   Fifteen placebo-controlled treatment groups with baseline and ending measures were analyzed...The results of this meta-analysis suggest that neither soy foods nor isoflavone supplements alter measures of bioavailable T concentrations in men.
   Fertil Steril, 2010 · checked 2026-09-24
9. sybn-r4-11 · regulatory decision
   POL-HC-6260 Soy protein Soy protein has been shown to lower/reduce blood cholesterol; blood cholesterol lowering may reduce the risk of (coronary) heart disease -/- Non-authorised
   EU Register on nutrition and health claims, claim POL-HC-6260, snapshot 2026-09-21 · checked 2026-09-24
10. sybn-r4-03 · systematic review and meta-analysis of RCTs
    Eighteen trials with a total of 522 participants were included in this systematic review...The findings of the meta-analysis showed a potential protective impact of soy protein intake on hyperlipidemia and proteinuria in CKD patients.
    Int Urol Nephrol, 2025 · checked 2026-09-24
11. sybn-r4-05 · network meta-analysis of RCTs
    Sixty RCTs involving 6284 participants (mean age: 54.2 years) were included. At the food level, no dairy or soy interventions significantly improved outcomes versus control, although milk + yogurt ranked numerically highest based on SUCRA values.
    Nutrients, 2025 · checked 2026-09-24
12. sybn-r4-12 · regulatory proposal
    The proposed rule would revoke the regulation authorizing the use of a health claim regarding the relationship between soy protein and risk of coronary heart disease
    Food Labeling: Health Claims; Soy Protein and Coronary Heart Disease, proposed rule, Federal Register, 2017 · checked 2026-09-24
13. sybn-r4-13 · umbrella review
    We identified 23 meta-analyses of observational research with 39 unique health outcomes and one meta-analysis of randomized controlled trials with four unique outcomes...However, it is important to note that the benefits of fermented soy products were inconclusive, and even some studies indicated their possible adverse effects.
    Mol Nutr Food Res, 2026 · checked 2026-09-24