Soy isoflavones

What it is

Soy isoflavones are phenolic compounds from the soybean, Glycine max, with a chemical structure similar to that of estradiol. The three main ones are genistein, daidzein and glycitein. Soybeans contain roughly 1.2 to 4.2 mg of these compounds per gram of dry weight, less concentrated than red clover. Daidzein is converted by gut bacteria in some people into a further compound, equol, and that conversion is a major source of person-to-person difference.

What the research says

Soy isoflavones are sold mainly for menopausal symptoms and bone health, and the evidence is genuinely split by outcome. Pooled randomised trials show small average improvements in total cholesterol and HDL, and a small increase in bone mineral density of about 0.4% to 1.9% depending on the site. For hot flushes, the picture is worse than the marketing suggests: a 2025 meta-analysis found no significant effect on hot flushes, sweating, insomnia or vasomotor symptoms, and a systematic review of trials in women with or at risk of breast cancer found five randomised trials showing no significant reduction versus placebo. Reported harms are mostly mild and digestive, along with allergic reactions in people allergic to soy. Case reports suggest soy may interfere with levothyroxine absorption, though a randomised crossover study did not find it. EFSA could not derive a safe intake level for isoflavone supplements, while finding no indication of adverse effects on the breast or uterus in post-menopausal women at the doses reviewed.

Evidence grade: Limited evidence.

What goes wrong

The same review flagged that the absence of long-term trials means serious adverse effects in breast cancer survivors could have been missed. (Source 1)

• Systematic review, Low certainty.
• Size: 40 RCTs.
• Who: breast cancer survivors.
• How long: most RCTs under 2 years.
• Result: no rates; a stated evidence gap on long-term safety, including possible interaction with tamoxifen.
• Funding: Canadian Institutes of Health Research.

This is a particularly important shortcoming because RCTs of under 2 years duration are unlikely to reveal any serious adverse effects in breast cancer survivors, including possible interactions between soy and tamoxifen.

Adverse effects of soy isoflavones reported in trials are mainly gastrointestinal, with allergic skin reactions in people allergic to soy. (Source 2)

• Meta-analysis, Low certainty.
• Size: twelve studies in the review.
• Who: perimenopausal women taking soy isoflavones.
• How long: varied.
• Result: no rates given; nausea, abdominal pain or diarrhoea, and skin itching, redness and rashes in soy-allergic individuals.
• Funding: institutional grants.

some individuals may experience gastrointestinal discomfort such as nausea, abdominal pain, or diarrhea after consuming soy isoflavones. Allergic reactions, including skin itching, redness, and rashes, may also occur among individuals allergic to soy products.

A systematic review of isoflavone supplements described side effects as mild and mostly gastrointestinal, and reported benign non-significant breast changes in one 12-week study. (Source 3)

• Systematic review, Low certainty.
• Size: 95 abstracts screened, 68 full-text articles reviewed.
• Who: menopausal women.
• How long: studies of varying length, including one 12-week study of breast changes.
• Result: no rates given; changes did not warrant intervention or follow-up for at least three months after the study.
• Funding: not stated.

Although soy isoflavones induced some benign clinical and non-significant ultrasonographic changes of the breast in one 12-week study, none of these changes warranted intervention or follow-up for at least three months after the study.

A narrative review of soy and levothyroxine found that case reports suggest possible interference with tablet absorption while the only randomised crossover study found none. (Source 4)

• Expert review, not systematic, Very low certainty.
• Size: case reports, case series and original studies published to 30 November 2022.
• Who: people taking levothyroxine, including those on liquid formulations.
• How long: not applicable.
• Result: no pooled estimate; a prospective randomised cross-over study showed no differences in levothyroxine absorption when taken with soy isoflavones.
• Funding: not stated.

Few data, mainly case reports, seemed to suggest a possible interference of soy products on L-T4 tablets absorption. However, the only prospective randomized cross-over study showed no differences in L-T4 absorption when L-T4 and soy isoflavones were assumed concomitantly.

EFSA's 2015 risk assessment could not derive a safe intake level for isoflavone food supplements, while finding no indication of adverse effects on the mammary gland at the doses and durations reviewed. (Source 5)

• Official position, Certainty not rated.
• Size: not applicable.
• Who: peri- and post-menopausal women without a current diagnosis or history of oestrogen-dependent breast or uterine cancer.
• How long: the doses and durations considered in the opinion.
• Result: no health-based guidance value or safe intake level could be derived; no indication of adverse mammary or uterine effects at the doses reviewed.
• Funding: European Union agency.

it was not possible to derive a single health-based guidance value or a safe intake level for food supplements containing isoflavones

What the evidence supports

A 2025 meta-analysis found soy isoflavones improved overall menopausal symptom scores and several specific symptoms, with small effect sizes. (Source 2)

• Meta-analysis, Low certainty.
• Size: seven studies, 533 participants for the overall menopausal symptom outcome; twelve articles in total.
• Who: perimenopausal women.
• How long: varied between trials; not pooled in the abstract.
• Result: menopausal symptoms Hedges' g = −0.25 (95% CI −0.42 to −0.08, p = 0.00); headache g = −0.38 (95% CI −0.60 to −0.17); psychosocial symptoms g = −0.29 (95% CI −0.48 to −0.10); palpitation g = −0.42 (95% CI −0.63 to −0.22); depression g = −0.72 (95% CI −1.17 to −0.28)
• Funding: supported by Inner Mongolia Medical University and regional government grants (institutional, not industry)

soy isoflavones were effective for treating menopausal symptoms (seven studies, 533 participants, Hedges' g = −0.25, 95% CI [−0.42 to −0.08], p= 0.00)

That review found no evidence of harm from soy with respect to breast cancer risk or recurrence, and reported that human trials show soy does not raise circulating estradiol. (Source 1)

• Systematic review, Low certainty.
• Size: 131 articles including 80 observational studies.
• Who: women with or at risk of breast cancer, including women taking tamoxifen.
• How long: long-term observational data.
• Result: no pooled estimate; the review could not pool risk data because of heterogeneity.
• Funding: Canadian Institutes of Health Research.

There is lack of evidence showing harm from use of soy with respect to risk of breast cancer or recurrence, based on long term observational data.

A meta-analysis of eighteen randomised trials found soy isoflavones produced small statistically significant gains in bone mineral density at three sites. (Source 6)

• Meta-analysis, Low certainty.
• Size: eighteen randomised controlled trials.
• Who: postmenopausal women.
• How long: 6 to 24 months.
• Result: lumbar spine WMD 1.63% (95% CI 0.51 to 2.75), p = 0004 as printed; femoral neck WMD 1.87% (95% CI 0.14 to 3.60), p = 0.034; total hip WMD 0.39% (95% CI 0.08 to 0.69), p = 0.013.
• Funding: This research received no external funding.

Limit of this finding: The published abstract prints the lumbar spine result as "p = 0004", which is not a valid p-value because a decimal point is missing. The quote above reproduces the journal's text exactly. The figure is almost certainly 0.004, so the lumbar spine gain was almost certainly statistically significant, but "0004" should not be read as a number in its own right and nothing about the size of the effect follows from it. The femoral neck and total hip p-values in the same sentence (0.034 and 0.013) are printed normally.

daily intake of 106 (range, 40–300) mg of isoflavones for 6–24 months moderately but statistically significantly positively affects BMD, compared with controls: lumbar spine WMD = 1.63 (95% CI: 0.51 to 2.75)%, p = 0004; femoral neck WMD = 1.87 (95% CI: 0.14 to 3.60)%, p = 0.034; and total hip WMD = 0.39 (95% CI: 0.08 to 0.69)%, p = 0.013.

A separate meta-analysis of 42 phytoestrogen trials also found soy protein lowered total and LDL cholesterol and raised HDL slightly in postmenopausal women. (Source 7)

• Meta-analysis, Low certainty.
• Size: 42 randomised controlled trials across flaxseed, soy and red clover.
• Who: postmenopausal women.
• How long: varied; not pooled in the abstract.
• Result: total cholesterol WMD −0.15 (95% CI −0.25 to 0.05), p = 0.0048; LDL-C WMD −0.15 (95% CI −0.25 to 0.05), p = 0.0067; HDL-C WMD 0.05 (95% CI 0.02 to 0.08), p = 0.0034.
• Funding: not stated in the accessible content.

Statistical analysis showed a significant decrease in TC levels: WMD = −0.15; 95% CI: −0.25–0.05; p = 0.0048, LDL-C levels: WMD = −0.15; 95% CI: −0.25–0.05; p = 0.0067, and a significant increase in HDL-C levels: WMD = 0.05; 95% CI: 0.02–0.08; p = 0.0034.

What the evidence does not support

The same meta-analysis found no significant effect of soy isoflavones on hot flushes, sweating, insomnia or vasomotor symptoms. (Source 2)

• Meta-analysis, Low certainty.
• Size: twelve studies; subgroup numbers vary by symptom.
• Who: perimenopausal women.
• How long: varied between trials.
• Result: no significant treatment effect for paresthesia, fatigue, physical symptoms, hot flushes, excessive sweating, insomnia or vasomotor symptoms.
• Funding: institutional grants from Inner Mongolia Medical University and regional government.

but no significant treatment effect on paresthesia symptoms, fatigue symptoms, physical symptoms, hot flushes, excessive sweating, insomnia, and vasomotor symptoms was observed

A systematic review of soy and red clover in women with or at risk of breast cancer found that five randomised trials of soy for hot flashes showed no significant reduction against placebo. (Source 1)

• Systematic review, Moderate certainty.
• Size: 131 articles: 40 RCTs, 11 uncontrolled trials and 80 observational studies, from 4179 records.
• Who: patients with or at risk of breast cancer.
• How long: inception to March 2013; trial durations varied.
• Result: five RCTs showed no significant reductions in hot flashes compared to placebo.
• Funding: Knowledge Synthesis grant from the Canadian Institutes of Health Research; funders had no role.

Five RCTs reported on the efficacy of soy for hot flashes, showing no significant reductions in hot flashes compared to placebo.

That review concluded any interference of soy with levothyroxine absorption is likely to have little clinical impact and that the question remains unresolved. (Source 4)

• Expert review, not systematic, Very low certainty.
• Size: not stated.
• Who: people taking levothyroxine.
• How long: not applicable.
• Result: no numbers; a qualitative conclusion.
• Funding: not stated.

The inference of soy products on L-T4 absorption, if present, seems to have little clinical impact.

Reviews of the equol hypothesis note that some studies of equol-producing ability have reported null or adverse effects rather than benefit. (Source 8)

• Expert review, not systematic, Low certainty.
• Size: not stated.
• Who: adults in observational and intervention studies.
• How long: not applicable.
• Result: no pooled estimate; mixed direction across studies, with clinical intervention trials of S-equol described as still needed.
• Funding: not stated.

Although some observational and intervention studies suggest that the ability to produce equol is associated with reduced risk of breast and prostate cancer, CVD, improved bone health and reduced incidence of hot flushes, others have reported null or adverse effects.

Where the evidence is mixed

The authors of that meta-analysis said their own findings should be read cautiously because of small samples, wide confidence intervals and inconsistent methods. (Source 2)

• Meta-analysis, Low certainty.
• Size: twelve studies.
• Who: perimenopausal women.
• How long: varied.
• Result: no revised estimate; a stated limitation on certainty.
• Funding: institutional grants.

the small effect sizes for many outcomes, wide confidence intervals, and methodological inconsistencies among the included studies limit the certainty of these findings

A meta-analysis of randomised trials in 2305 postmenopausal women found small reductions in total cholesterol and a borderline rise in HDL, with LDL and triglyceride changes not reaching significance. (Source 9)

• Meta-analysis, Low certainty.
• Size: 2305 postmenopausal women.
• Who: postmenopausal women.
• How long: varied between trials.
• Result: total cholesterol −0.12 mmol/L (95% CI −0.21, −0.03), p = 0.01; HDL +0.03 mmol/L (95% CI 0.00, 0.06), p = 0.05; LDL −0.05 mmol/L (95% CI −0.11, 0.01), p = 0.08; triacylglycerols −0.07 mmol/L (95% CI −0.14, 0.00), p = 0.06.
• Funding: no funding statement located.

Limit of this finding: The source's own sentence is internally inconsistent. It opens by describing "statistically significant decreases" and then lists LDL-cholesterol (p = 0.08) and triacylglycerols (p = 0.06) inside the same sentence, although neither reached the 0.05 level the paper itself uses, and it describes HDL-cholesterol as increased when the reported interval runs down to 0.00 with p = 0.05 exactly. A reader should treat the fall in total cholesterol as the one clearly significant result, and should not conclude that LDL-cholesterol or triglycerides were lowered, or that the HDL rise is firmly established.

Changes in the lipid profile showed statistically significant decreases of total cholesterol by −0.12 (95% CI: −0.21, −0.03) mmol/L, −4.64 (95% CI: −8.12, −1.16) mg/dL, p = 0.01

The same review stated that high heterogeneity across the soy isoflavone literature makes firm conclusions difficult, and that studies of isoflavones and cognition had serious methodological flaws. (Source 3)

• Systematic review, Low certainty.
• Size: 68 full-text articles; a cited systematic review of twelve RCTs on cognition.
• Who: postmenopausal women.
• How long: varied.
• Result: no pooled estimate; a stated limitation on the whole body of evidence.
• Funding: not stated.

The common finding in all of the research included in this review is that past studies have shown high heterogeneity, making it difficult to draw conclusions.

Roughly 42% of a Japanese cohort of 466 adults were equol producers, so a large minority of people cannot make the metabolite that some of the proposed benefits are attributed to. (Source 10)

• Survey study, Moderate certainty.
• Size: 466 participants: 292 men and 174 women aged 22 to 88.
• Who: Japanese men and women.
• How long: single measurement of serum and urine.
• Result: 195 of 466 (42%) were equol producers; 47% of women and 39% of men; producers defined by log(equol/daidzein) of −1.42 or higher.
• Funding: not stated.

Among 466 participants, 195 were equol producers (42%). The proportion of equol producers in women was 47%, whereas that in men was 39%.

Population reviews put equol-producing ability at about 30% of Western adults and 60% of Asian adults, and note that the health consequences of producing it are not settled. (Source 8)

• Expert review, not systematic, Low certainty.
• Size: not stated.
• Who: Western and Asian adults given a soya challenge.
• How long: not applicable.
• Result: approximately 30% of Western adults and 60% of Asian adults produce equol; observational and intervention studies report both benefits and null or adverse effects.
• Funding: not stated.

approximately 30% of adults among western populations and 60% of Asian adults have the ability to produce equol from daidzein following a soya challenge.

Where the research disagrees

Whether soy isoflavones reduce hot flushes

• Authors of a 2019 systematic review of isoflavone supplements (Nutrients), systematic review without pooling: We found that isoflavones reduce hot flashes even accounting for placebo effect, attenuate lumbar spine bone mineral density (BMD) loss, show beneficial effects on systolic blood pressure during early menopause, and improve glycemic control in vitro. (Source 3)
• Authors of a 2025 meta-analysis in perimenopausal women (PeerJ), meta-analysis of randomised trials: but no significant treatment effect on paresthesia symptoms, fatigue symptoms, physical symptoms, hot flushes, excessive sweating, insomnia, and vasomotor symptoms was observed (Source 2)
• Authors of a 2013 systematic review in women with or at risk of breast cancer (PLOS ONE), systematic review of randomised trials: Five RCTs reported on the efficacy of soy for hot flashes, showing no significant reductions in hot flashes compared to placebo. (Source 1)

Whether soy interferes with levothyroxine

• Authors of a 2023 narrative review in Endocrine Practice, case reports summarised in a narrative review: Few data, mainly case reports, seemed to suggest a possible interference of soy products on L-T4 tablets absorption. (Source 4)
• The same authors, on the randomised evidence, prospective randomised crossover study: However, the only prospective randomized cross-over study showed no differences in L-T4 absorption when L-T4 and soy isoflavones were assumed concomitantly. (Source 4)

How much

• Reference intake: No reference intake exists for soy isoflavones: they are not essential nutrients. EFSA's 2015 opinion states that no single health-based guidance value or safe intake level could be derived for isoflavone food supplements. (Source 5)
• Upper limit: No upper limit or ADI has been set. EFSA's 2015 opinion instead reported no indication of adverse effects on the mammary gland in post-menopausal women at the doses and durations it reviewed, and limited its conclusions to women without a current diagnosis or history of oestrogen-dependent breast or uterine cancer. (Source 5)
• Studied: Across eighteen randomised bone trials, a daily intake of 106 mg of isoflavones (range 40 to 300 mg) for 6 to 24 months. (Source 6)
• Studied: 60 mg of isoflavones daily for 12 weeks in a prospective study of 51 healthy postmenopausal women; and a supplement containing 72 mg of isoflavones from soy beans and red clover for 6 months. (Source 3)

A common belief, and what the research shows

The belief: Soy isoflavones act like oestrogen in the body, so they raise breast cancer risk and must be avoided by anyone with a hormone-sensitive condition.

What the research shows: The human evidence points the other way on both halves. A systematic review of 131 articles concluded that "Soy does not have estrogenic effects in humans." and that "Human trials show that soy does not increase circulating estradiol or affect estrogen-responsive target tissues.", adding that "There is lack of evidence showing harm from use of soy with respect to risk of breast cancer or recurrence, based on long term observational data." EFSA's 2015 assessment found "no indication for adverse effects on the mammary gland in post-menopausal women from isoflavones when taken in doses and for durations as described above". The honest caveats are that the same review noted "There is a lack of long term interventional data assessing cancer risk.", that EFSA restricted its conclusions to "post-menopausal women without a current diagnosis or history of oestrogen-dependent breast or uterine cancer", and that EFSA could not set a safe intake level at all.

Questions and answers

What is it?

Soy isoflavones are plant phenolic compounds from the soybean, Glycine max. The three main ones are genistein, daidzein and glycitein. Their chemical structure resembles that of the human oestrogen estradiol, which is why they are described as phytoestrogens. They are found across the legume family, in soy, clovers, alfalfa and beans. (Source 11)

What does it do in the body?

Isoflavones bind weakly at oestrogen receptors, but a key step happens in the gut: bacteria convert daidzein into a further compound called equol, and only some people can do this. Much of the biological activity that researchers attribute to soy is attributed to equol rather than to daidzein itself, which is why two people eating the same soy can end up with different circulating compounds. (Source 10)

Is it good or bad for you?

It depends entirely on the outcome. Pooled trials show small favourable changes in cholesterol and bone density, and systematic reviews found no evidence of harm to breast cancer risk. But for the symptom soy is most often sold for, hot flushes, a 2025 meta-analysis found no significant effect. Side effects in trials are mild and digestive, and soy-allergic people can react. (Source 2)

How do you get more of it?

Isoflavones come from soy foods and from extract supplements. Bone trials pooled in 2022 used an average of 106 mg a day, ranging from 40 to 300 mg, over 6 to 24 months. Other trials used 60 mg a day for 12 weeks, or soy nuts substituted for other protein. This describes what trials gave people, not a recommendation. (Source 6)

If it is harmful, what reduces it?

Isoflavones are not stored, so reducing them means eating less soy or stopping the supplement. That matters for two groups: people who get digestive symptoms from isoflavone supplements, and people with soy allergy who can develop skin reactions. Both effects reported in trials were mild and tied to current intake rather than accumulation. (Source 2)

Why might someone be low in it or missing it?

Soy isoflavones are not essential nutrients, so nobody is deficient in them. Two things do vary. One is diet: intake is low where soy foods are not eaten. The other is biology: only about 30% of Western adults and 60% of Asian adults can convert daidzein into equol, so many people never form the metabolite that several proposed benefits are pinned on. (Source 8)

Which whole foods contain it or feed it?

Soybeans and soy foods are the main dietary source, holding roughly 1.2 to 4.2 mg of isoflavones per gram of dry weight. Other legumes in the same family, including red and white clover, alfalfa and beans, also contain them, and red clover is considerably more concentrated at 10 to 25 mg per gram dry weight. (Source 11)

What happens if you do not have it?

There is no deficiency disease from not eating soy isoflavones. Whether lacking them matters at all is unsettled even for the equol pathway: reviews note that while some studies link equol-producing ability to lower risk of several diseases, others have reported null or adverse effects, and intervention trials are still called for. (Source 8)

How can you test for it?

There is no test of isoflavone status, but equol-producer status can be measured. Research laboratories measure equol and daidzein in serum and urine by LC-MS/MS and classify producers using a cut-off on the urinary ratio. Serum and urine agreed well in producers but not in non-producers, so the method depends on which sample and cut-off are used, and it is a research classification rather than a validated clinical test. (Source 10)

References

1. PLOS ONE. Soy, Red Clover, and Isoflavones and Breast Cancer: A Systematic Review. 2013. PMID 24312387, DOI 10.1371/journal.pone.0081968. Read the source
2. PeerJ. Effects of soy isoflavones on menopausal symptoms in perimenopausal women: a systematic review and meta-analysis. 2025. DOI 10.7717/peerj.19715. Read the source
3. Nutrients. Isoflavone Supplements for Menopausal Women: A Systematic Review. 2019. DOI 10.3390/nu11112649. Read the source
4. Endocrine Practice. Interference or Noninterference Between Soy and Levothyroxine: That Is the Question. A Narrative Review of Literature. 2023. DOI 10.1016/j.eprac.2023.08.009. Read the source
5. EFSA Journal (European Food Safety Authority), 2015;13(10):4246; adopted 21 October 2015. Risk assessment for peri- and post-menopausal women taking food supplements containing isolated isoflavones. 2015. DOI 10.2903/j.efsa.2015.4246. Read the source
6. Journal of Clinical Medicine. The Role of Soy Isoflavones in the Prevention of Bone Loss in Postmenopausal Women: A Systematic Review with Meta-Analysis of Randomized Controlled Trials. 2022. DOI 10.3390/jcm11164676. Read the source
7. Nutrients. Role of Phytoestrogen-Rich Bioactive Substances (Linum usitatissimum L., Glycine max L., Trifolium pratense L.) in Cardiovascular Disease Prevention in Postmenopausal Women: A Systematic Review and Meta-Analysis. 2022. DOI 10.3390/nu14122467. Read the source
8. Proceedings of the Nutrition Society. Is equol production beneficial to health?. 2011. Read the source
9. Nutrients. Effects of Soy Protein Containing of Isoflavones and Isoflavones Extract on Plasma Lipid Profile in Postmenopausal Women as a Potential Prevention Factor in Cardiovascular Diseases: Systematic Review and Meta-Analysis of Randomized Controlled Trials. 2021. DOI 10.3390/nu13082531. Read the source
10. PLOS ONE. Comparison of blood and urine concentrations of equol by LC‒MS/MS method and factors associated with equol production in 466 Japanese men and women. 2024. DOI 10.1371/journal.pone.0288946. Read the source
11. Nutrients. Scientific Evidence Supporting the Beneficial Effects of Isoflavones on Human Health. 2020. DOI 10.3390/nu12123853. Read the source