Soy, hormones and the questions people actually ask

Read this first

This page is a summary of published research for general information. It is not medical advice and nothing here is personalised to you. If you have a thyroid condition, a soy allergy, a hormone-sensitive illness, or you are currently in treatment for anything, that conversation belongs with your doctor rather than with a cooking website.

The aim below is to be accurate rather than reassuring. Where the evidence is strong it says so, and where it is limited or contested it says that too.

What isoflavones are

Soybeans contain compounds called isoflavones, which are classified as phytoestrogens. They are structurally similar enough to the hormone estradiol to bind to human estrogen receptors, and they bind considerably more weakly than the hormone itself does.

That structural similarity is the entire basis of every concern raised about soy. It is a real fact and not a myth. What is contested is what follows from it in the body, and that is a question the laboratory chemistry cannot answer on its own.

A block of tofu contains a meaningful amount of isoflavones, though the exact figure varies with the bean, the process and the firmness. Isolated isoflavone supplements deliver them at concentrations food does not, and the two exposures should not be treated as interchangeable.

Does soy affect testosterone in men

This is the most persistent claim, and it is the one where the clinical evidence is strongest and most consistent.

A meta-analysis published in Reproductive Toxicology in 2021 by Reed and colleagues pooled dozens of clinical studies and found no effect of soy protein or isoflavone intake on total testosterone, free testosterone, estradiol or estrone in men. Sub-analyzes by dose and by study duration were also null. An earlier meta-analysis in 2010 reached the same conclusion on bioavailable testosterone.

A dose-response meta-analysis published in Food Frontiers in 2025 by Rajaie and colleagues agreed with the earlier work on testosterone and free testosterone, but reported findings on estradiol that differed from the previous analyzes. So the testosterone question looks settled and the estradiol question is less so, which is worth stating plainly rather than rounding off.

The claims that circulate about soy feminising men trace largely to a small number of case reports involving extremely high intakes, well outside anything a person eating tofu twice a week would reach.

Soy and breast cancer

This is where most of the genuine fear sits, and the picture has shifted considerably over the last two decades.

A systematic review and meta-analysis published in JNCI Cancer Spectrum in January 2024, co-directed by investigators at the Johns Hopkins Kimmel Cancer Center, examined observational studies of soy and other plant compounds in breast cancer survivors. Soy isoflavone intake was associated with roughly a quarter lower risk of recurrence, with the association strongest among postmenopausal and estrogen receptor positive survivors. The dose-response analysis found the largest reduction around 60 mg a day, with no additional benefit above that.

The American Cancer Society's 2022 nutrition guideline for cancer survivors concluded that soy food consumption before diagnosis is associated with lower overall mortality.

A separate 2024 meta-analysis of randomised trials in Advances in Nutrition examined whether isoflavones behave like estrogen on measures of estrogenicity in postmenopausal women, and found that they did not on the outcomes assessed. The authors were careful to note this does not rule out estrogen-like effects on outcomes they did not measure or in other groups.

Where the evidence is weaker than headlines suggest

Most of the cancer findings come from observational studies rather than randomised trials. Observational research can show associations and cannot establish cause. People who eat more soy differ from people who eat less in many ways, and statistical adjustment only goes so far.

Much of the strongest data comes from East Asian populations with lifelong high soy intake, and it is not obvious that the same findings transfer to someone who starts eating tofu at forty.

Conflicts of interest exist in this literature and are worth knowing about. Some prominent authors of soy meta-analyzes have consulted for soy food companies, which is disclosed in their publications. That does not invalidate the work, and it is a reason to weight independent replications more heavily than any single research group.

Soy foods and isolated isoflavone supplements are different exposures. Findings about tofu, tempeh and soy milk do not automatically apply to concentrated extracts, and most of the reassuring data is about food.

The other questions

Thyroid. Soy can interfere with the absorption of thyroid hormone medication, which is a practical timing issue rather than a reason to avoid soy. Anyone taking thyroid medication should raise it with their prescriber. There is also longstanding discussion about soy and thyroid function in people with iodine deficiency, which is another conversation for a doctor.

Allergy. Soy is a common food allergen and this is not a question of degree. Anyone with a diagnosed soy allergy should not eat tofu.

Processing. Tofu is made from soybeans, water and a coagulant, usually a calcium or magnesium salt. It is a minimally processed food by any reasonable definition, and the process has been essentially unchanged for centuries. Whatever one thinks about ultra-processed foods, tofu is not a useful example of one.

Children and infants. Genuinely a separate question with a different literature, and one for a paediatrician rather than a summary page.

How to look into it yourself

Every study named above is findable by author and journal. Reading the abstracts directly is more useful than reading anyone's summary, including this one, because the abstracts state their own limitations in a way that secondary coverage rarely does.

When you encounter a strong claim in either direction, three questions sort most of it: was this a clinical trial or an observational study, was the exposure soy food or an isolated supplement, and does the dose resemble anything a person would actually eat.

Corrections to this page are welcome and get made. Send them through the contact page with a source, and they will be acted on.

For the rest of the background section, the nutrition page covers what is in a block by type, and the learn hub links everything else.