Soy

Soy foods are a good protein source with a modest cholesterol-lowering effect, and the fears around them mostly fail on the evidence. In large observational studies soy eaters have no more breast cancer, and women who eat soy after a breast cancer diagnosis have fewer recurrences; in trials, soy does not lower men's testosterone or upset normal thyroid hormone levels. The cautions are narrow: isoflavone pills, and people whose thyroid is already struggling.

Eat tofu, tempeh, edamame and unsweetened soy milk freely as part of a legume-rich diet, including after breast cancer. Skip isoflavone supplements unless a clinician suggests them; with a mildly underactive thyroid, recheck TSH if you eat a lot of soy; and if you take levothyroxine for an underactive thyroid, take it at a different time from soy.

What it is

Soybeans are legumes, eaten whole as edamame, curdled into tofu, fermented into tempeh, miso and natto, or pressed into soy milk. They also carry isoflavones, plant compounds shaped enough like the hormone estrogen to bind weakly to its receptors, which is the root of every scare about soy. Natto, the fermented Japanese soybean, has its own evidence and its own page: Natto. Beans, lentils and soy as a food group are covered under Legumes and soy.

One caveat about the research below: Mark Messina, a co-author of several of the meta-analyses on this page (on cholesterol, men's hormones and hot flushes) and of the soy-and-muscle analysis on Legumes and soy, was affiliated with the Soy Nutrition Institute when the cholesterol analysis was published.^[1] The large Chinese breast cancer cohort and the thyroid review were done without him.

The heart: a small cholesterol drop, and associations with fewer deaths

Soy protein lowers low-density lipoprotein (LDL) cholesterol, but only a little. When researchers pooled the 46 trials the US Food and Drug Administration (FDA) had assembled while considering whether to withdraw soy's heart-health label claim (41 of them reported LDL), a median 25 g of soy protein a day for about six weeks lowered LDL by about 0.12 mmol/L (4.76 mg/dL), roughly 3–4%, compared with other proteins.^[2] Most of the trials pointed the same way, though few were significant on their own.

Across 23 cohort studies of about 330,000 people, the biggest soy eaters had about 15% fewer cardiovascular deaths and about 12% fewer cancer deaths than the smallest. A link with deaths from any cause showed up only in some subgroups, mainly the higher-quality studies, and soy protein on its own was not linked to overall or cardiovascular death.^[3] These are observational: in many of the cohorts soy eaters differ from non-eaters in other ways too. In the largest Japanese study, it was fermented soy, not plain soy, that went with lower mortality; see Natto.

Breast cancer: no harm in the data, and fewer recurrences in survivors

The worry is that estrogen-like compounds might feed estrogen-sensitive tumours. The human data don't show it.

Risk of getting it. Among more than 300,000 Chinese women followed for about ten years, the women eating the most soy (about 19 mg of isoflavones a day) got breast cancer at the same rate as those eating the least (about 4.5 mg). Pooled with earlier cohorts, each extra 10 mg of isoflavones a day went with about 3% lower risk.^[4] An earlier pooling found the lower risk only in Asian populations (about 24% lower) and no association either way in Western ones, where intakes are far smaller.^[5]

After a diagnosis. In 9,514 breast cancer survivors in the US and China followed for about seven years, women eating at least 10 mg of isoflavones a day had about 25% fewer recurrences. Their lower breast-cancer death rate (about 17%) could have been chance.^[6] A 2024 meta-analysis found the same: about 26% fewer recurrences, including in women with estrogen-receptor-positive tumours, the ones the worry was about.^[7] All of this is observational and describes women who ate soy foods as a habit. Whether starting a high intake, or taking isoflavone pills, after diagnosis does the same has not been tested.

Men: no effect on hormones, a weaker case for the prostate

Testosterone and "feminisation". Several case reports describe men with feminising changes, such as lower testosterone and higher estrogen, blamed on soy. Controlled studies don't bear that out. Pooling 41 clinical studies, with testosterone measured in 1,753 men, neither soy protein nor isoflavones changed total or free testosterone, estradiol or estrone, whatever the dose or duration.^[8] For muscle, see Legumes and soy: with resistance training, soy protein built muscle and strength as well as animal protein did.

Prostate cancer. Pooling the 16 studies of total soy intake in a review of 30, soy eaters had about 29% lower prostate cancer risk. But most of that came from case-control studies, which ask people with and without cancer to recall their past diet and are prone to recall bias. In the sturdier cohort studies the reduction was about 10%, and soy was not linked to advanced prostate cancer.^[9] Treat it as a possible small benefit, not a reason to eat soy.

The thyroid: hormone levels hold steady, with two cautions

In 18 randomised trials, soy did not change the thyroid hormones themselves. Thyroid-stimulating hormone (TSH), the pituitary signal that rises when the thyroid is struggling, went up by about 0.25 mIU/L, a change whose clinical meaning, if any, is unclear.^[10]

Subclinical hypothyroidism (a mildly high TSH with normal thyroid hormones) is the one group with a signal, and it is mixed. In a trial of 60 people with it, six women progressed to overt hypothyroidism on a supplement giving 16 mg of isoflavones a day — a dose the authors chose as representative of a vegetarian diet — about three times the expected rate.^[11] The same group's follow-up, in 44 people at a much higher 66 mg a day, found no increase in thyroid failure and no change in thyroid tests, though two of the 44 still progressed.^[12] Both tested supplements, but the first at a dose food can supply.

Levothyroxine. A systematic review of food interactions found limited evidence that soy products, like coffee, fibre, calcium and iron, reduce absorption of levothyroxine, the standard thyroid replacement pill, and that keeping a time gap between the pill and food is an effective fix.^[13] The interaction is mostly case reports, and the one randomised crossover study, which tested soy isoflavones, found no effect, so it is probably small.^[14] More on testing and treatment is under Thyroid.

Hot flushes: a modest effect from pills, not recommended by the specialists

Isoflavone supplements do something for menopausal hot flushes. Pooling 17 placebo-controlled trials, a median 54 mg a day for six weeks to a year cut flush frequency by about 21% and severity by about 26%, with results varying a lot between trials.^[15] Hormone therapy remains the most effective treatment, and the North American Menopause Society's (now The Menopause Society) 2023 review of non-hormonal options lists soy foods and soy extracts among the treatments it does not recommend.^[16] The options that work are on Menopausal hormone therapy and Neurokinin blockers.

Who should be careful

People taking levothyroxine: take it apart from soy foods and soy milk, as with coffee and calcium.
People with subclinical hypothyroidism: avoid isoflavone supplements, and if you eat a lot of soy, have your TSH rechecked.
Anyone on a classic monoamine-oxidase inhibitor (MAOI) antidepressant, or with histamine intolerance: fermented soy, natto included, is among the richer food sources of tyramine and histamine; see Natto.

What to do

Use soy as one of your everyday proteins — tofu, tempeh, edamame, unsweetened soy milk.
After breast cancer, there is no reason to avoid soy foods. Don't add isoflavone pills.
Men need not worry about testosterone.
Prefer food to isoflavone supplements. Their hot-flush benefit is modest, and the reassuring breast cancer data are about food.