Is soy dangerous to health?

Authorities do not consider it dangerous in itself, but they are cautious about exposure to isoflavones. In 2025, ANSES set toxicological reference values (0.02 mg/kg per day, and 0.01 for pregnant women or women of childbearing age and for prepubertal children) and advises against soy products in collective catering. This article does not replace medical advice.

Does soy increase the risk of breast cancer?

Meta-analyses of cohort studies reviewed show no increase in risk: relative risk of 0.99 for total isoflavone intake (16 studies, Clinical Nutrition 2019), and high consumption of soy foods is associated with a lower risk. These are observational studies. In 2005, ANSES recommended limited intake for people with a history of hormone-dependent cancer: this is a medical question.

Does soy lower men's testosterone?

No, according to a meta-analysis of 41 clinical studies (Reproductive Toxicology, 2021): neither soy nor isoflavones significantly affect testosterone, estradiol or estrone in adult men.

Is soy responsible for deforestation?

Partly, but mostly through livestock farming: according to Our World in Data (2018 data), around 77% of the world's soy feeds animals and 7% is eaten directly by humans. The direct role of soy in Amazon deforestation has declined since the 2006 moratorium, but indirect risks remain.

Soy: friend or foe? What the studies say

What is soy, and why is it debated?

Soy is a protein-rich legume, long consumed in Asia and increasingly in Europe as tofu, tempeh, drinks, edamame (fresh beans), soy sauce or textured proteins. It is also very present, as oil, lecithin or protein, in many industrial products.

The debate comes from a family of molecules it contains: theisoflavones, also calledphytoestrogens. Their structure resembles that of estradiol, a female hormone, and they bind to the same receptors, though with lower affinity. According to the 2005 report by Afssa (now ANSES), they behave as modulators of these receptors, which illustrates, according to the report, how hard their effects are to predict. Having hormone-like activity does not mean producing the same effects as a hormone, nor being dangerous at any dose.

Where is soy hiding on our plate?

Isoflavone levels vary enormously from one product to another. In an opinion published in 2025, ANSES notes for example that soy appetiser biscuits contain around100 times morethan soy sauce, and that soy desserts can vary twofold. Talking about "soy" in general therefore means little: an edamame bowl, a sauce and an industrial biscuit are not equivalent.

A useful clarification: in the classic Western diet, without soy products, isoflavone intake is very low, on the order of 26 micrograms per day on average in adults (18 in 3 to 15-year-olds) according to the 2005 report. Among soy-food consumers alone, it is 1,000 to 10,000 times higher (around 15 mg per day on average, on a small sample).

Soy and deforestation: which soy are we talking about?

The image of soy destroying the Amazon rainforest is partly founded, but it mainly concerns livestock farming. According to Our World in Data (2018 data from the USDA), around77% of the world's soyis used to feed farm animals, 37% of it for poultry. Only7%is eaten directly by humans (tofu, drinks, edamame, tempeh), and around 4% goes to biofuels and industrial uses.

Our World in Data also nuances the picture: the direct role of soy in Amazon deforestation has declined since the 2006 moratorium, and cattle pasture remains the dominant driver there. Indirect risks (displacement of livestock farming, expansion into the Cerrado) remain. In short, eating tofu does not have the same environmental weight as eating meat from animals fed on imported soy.

Soy and breast cancer: what do the studies say?

It is the most widespread fear, since breast cancer can be sensitive to hormones. Human data are nonetheless rather reassuring:

A meta-analysis of 16 cohort studies (11,169 cases among 648,913 participants, Clinical Nutrition, 2019) findsno associationbetween total isoflavone intake and breast cancer risk (relative risk of 0.99). High consumption of soy-based foods is, for its part, associated with a lower risk.
A 2022 meta-analysis (a more modest review, with no numerical estimate in the abstract) concludes that prospective studies show an inverse association between isoflavones and breast cancer, in women both before and after menopause.
The 2005 ANSES report notes that a reduced risk is observed in Asian women, but that this cannot be transposed to Western women, whose diets and intakes differ.

These areobservational studies: they show associations, not causes. For women who have had hormone-dependent cancer, ANSES recommended in 2005 a limited intake of phytoestrogens. This is a medical situation: only a health professional can say what to do in that case. EFSA's 2015 opinion, focused on isoflavone supplements in postmenopausal women, did not conclude to a harmful effect on the breast, uterus or thyroid at the doses studied either, while noting that it was not possible to set a single reference value (according to the opinion's summary records).

Soy, thyroid and testosterone

Thyroid.A meta-analysis of 18 randomised trials (2019) finds no effect of soy on the thyroid hormones fT3 and fT4, and only a very modest rise in TSH, whose clinical significance the authors consider uncertain. By contrast, the 2005 report listed peoplewith treated hypothyroidismamong the groups to consider, as interactions with treatments have been observed. Those concerned can discuss this with their doctor.

Testosterone.A meta-analysis of 41 clinical studies in men (Reproductive Toxicology, 2021) finds no significant effect of soy or isoflavones on testosterone, estradiol or estrone. The myth of soy that "feminises" men is not supported by these trials.

Infants and children: official caution

This is where authorities are most reserved. In its 2005 report, Afssa judged human studies on infants "very insufficient" and recommended limiting isoflavones in soy infant formulas to1 mg per litreof reconstituted milk, whereas those studied contained 32 to 47 mg/L. It classed foetuses, infants and young children among the at-risk groups.

Another point not to confuse: in 2013, ANSES reminded thatplant-based drinks, including soy ones, are not suitable for feeding infants under one year old, as they expose them to deficiencies. They replace neither breast milk nor an infant formula. We have not been able to verify the current position of the HAS or EFSA on soy infant formulas: for a baby, the question is one for the paediatrician.

Is there an official consumption limit?

Yes, recently. In its 2025 opinion, ANSES sets toxicological reference values, based on toxic effects on the reproductive system, of0.02 mg/kg body weight per dayof isoflavones for the general population, and0.01 mg/kg/dayfor pregnant women or women of childbearing age and for children before puberty. It estimates that 76% of 3 to 5-year-old children who consume soy foods exceed this threshold. It recommends not serving them incollective catering, whatever the age, and diversifying legumes.

These values are lower than the 1 mg/kg/day dose up to which the 2005 report observed no adverse effect of general toxicity, which targeted a different type of effect, and they have been disputed by the soy sector, which regrets the absence of a benefit-risk assessment. They do not target one food in particular but total exposure, which depends on the products chosen.

Allergy, processed products and origin

Allergy.Soy is one of the allergens that must be declared on the label. People with an allergy must avoid it, whatever its form.
Processed products.Plant steaks, soy desserts, flavoured drinks: many fall into the ultra-processed food category (NOVA 4), with long ingredient lists. Plain tofu has nothing in common with a plant-based nugget: the form matters more than the plant.
GMOs.EU regulations require labelling of food containing or produced from genetically modified organisms, with a 0.9% threshold for accidental presence. We have not been able to verify here the current share of GM soy in imports.

What to remember and look for on Yumiq

Soy contains isoflavones with hormone-like activity, hence a real but often exaggerated debate.
The meta-analyses reviewed show no effect on testosterone and no increase in breast cancer risk.
Health authorities remain cautious for infants, children and certain medical situations.
Levels vary from one product to another: the quantity and the form matter more than the word "soy".
Around 77% of the world's soy feeds animals: soy on the plate is not the main deforestation issue.

On Yumiq, the product page shows the number of additives detected (harmless, to monitor, or of concern). TheYumiq-Scorepenalises controversial additives assessed by EFSA and ANSES, and the Yumiq star requires the absence of any additive of concern. To go further on alternatives, seeCow's milk vs plant-based milkandAre plant proteins enough?.

Soy: friend or foe?