Is soy milk safe during pregnancy?

In the amounts people drink, we found nothing that shows harm, and nothing that tested it directly. No randomized trial has given soy milk to pregnant women. The firmest evidence comes from people who were not pregnant: a meta-analysis of 18 randomized trials found soy left thyroid hormones unchanged and raised TSH by about 0.25 mIU/L, a rise whose clinical meaning the authors call unclear. The NIH National Center for Complementary and Integrative Health flags soy in pregnancy only in amounts greater than those commonly found in foods.

Unsettled. The answer for pregnancy itself rests on cohorts and one case-control study, plus trials in women who were not pregnant. Nobody has published a randomized trial of soy foods in pregnancy, and we found no systematic review of soy intake in pregnancy either. So the honest wording is no sign of harm at food amounts. Proven safe would claim more than the studies show.

What the trials found

The worry about soy in pregnancy is that its isoflavones act like a weak estrogen. Trials outside pregnancy put a size on that. A meta-analysis of 47 trials in women, 11 of them in premenopausal women, found soy or isoflavones did not change estradiol, estrone or SHBG before menopause. FSH and LH fell by about 20 percent on a standardized scale, and the cycle lengthened by about a day.

For soy drink itself, a 2026 network meta-analysis of 14 randomized trials in 543 adults found that swapping 17 fl oz (500 mL) a day of cow milk for soy drink may lower LDL cholesterol by 18 mg/dL (95% CI -33 to -3.9; metric: 0.47 mmol/L, -0.85 to -0.10), on low certainty evidence. Nobody in those trials was pregnant, and we found no trial of the drink in pregnant women.

The one randomized trial in pregnancy tested soy protein in a diet. The drink was not part of it. In 68 Iranian women with gestational diabetes, six weeks of a diet with 35 percent of its protein from soy improved glucose control and was linked to fewer newborns with jaundice, in the authors' summary. It is one small trial, and this author group has had data integrity concerns raised about other papers. This paper is not in Retraction Watch, checked on 7 October 2026.

Birth defects, the question behind most searches

The concern started with hypospadias, a malformation of the penis. In the UK ALSPAC cohort of 7,928 boys, mothers who were vegetarian in pregnancy had an adjusted odds ratio of 4.99 (95% CI 2.10 to 11.88) for a son with hypospadias, from 51 cases. That study never measured soy, and phytoestrogens were the authors' hypothesis.

Later and larger studies did measure it, and the direction did not hold. In a Japanese cohort of 41,578 mothers of boys, the lowest isoflavone intake in early pregnancy carried an odds ratio of 2.8 (95% CI 1.4 to 5.8) for hypospadias, and the highest intake 0.9 (95% CI 0.4 to 2.4), from 51 cases. In a US case-control study of 1,250 boys with hypospadias and 3,118 controls, high total phytoestrogen intake had an odds ratio of 0.7, with an interval from 0.5 to 1.0 that reaches no effect. A 2019 meta-analysis of 19 observational studies of vegetarian diets in pregnancy could not reach a conclusion on hypospadias, fetal growth, anemia or gestational diabetes.

Gestational diabetes, an association in the other direction

In a Japanese cohort of 84,948 pregnant women, those in the top fifth of soy isoflavone intake had a relative risk of gestational diabetes of 0.82 (95% CI 0.70 to 0.95) against the bottom fifth. The link was with miso and natto, and soy milk was not analyzed separately. In a Chinese cohort of 1,495 pregnant women, the top third of soy food intake had a relative risk of 0.73 (95% CI 0.54 to 0.99) against no soy, at a median intake of only 2.9 g a day. These are associations and cannot show that soy prevents anything.

Who should be careful

Not for everyone. Soy allergy is the direct reason to avoid soy milk, pregnant or not. The NIH center notes that some people are allergic to soy, and US labels must declare it.

Iodine is the caution most likely to matter in practice. In the UK national diet survey, people who drank only milk alternatives had a median urinary iodine of 79 micrograms per liter, which the World Health Organization classes as deficient, against 132 in cow milk drinkers. Their iodine intake was 94 against 129 micrograms a day. Those drinks were mostly unfortified at the time, and the people were not pregnant. Pregnancy raises the need for iodine, so swapping cow milk for an unfortified soy drink removes a main source of it. Check the label for added iodine.

Thyroid is the other question pregnant women ask. In 505 women in early pregnancy in China, eating soy three or more times a week was not linked to thyroid dysfunction or thyroid autoimmunity. That study counted how often, not how much. The trial meta-analysis above found the TSH rise of about 0.25 mIU/L in people who were not pregnant. If you are pregnant and treated for an underactive thyroid, how much soy fits is a question for whoever manages your thyroid.

Supplements are a different matter from the drink. The NIH center says soy may be unsafe in pregnancy in amounts greater than those commonly found in foods, which leaves isoflavone tablets outside what the evidence above covers.

What expert bodies say

The NIH National Center for Complementary and Integrative Health, in its soy fact sheet updated April 2025, says the use of soy during pregnancy may be unsafe in amounts greater than those commonly found in foods. We found no position from the European Food Safety Authority on soy in pregnancy, because its 2015 isoflavone opinion covered only women around and after menopause. We could not reach NHS or Food Standards Agency guidance on soya drinks in pregnancy.

How we searched

What would change this answer

• Publication of NCT02806739, or any randomized trial of soy foods or soy milk in pregnant women with pregnancy outcomes measured.
• A meta-analysis of maternal soy or phytoestrogen intake and hypospadias. The three studies above point in different directions and have never been pooled.
• Iodine measured in pregnant women who drink plant milks, fortified and unfortified, rather than in the general population.

More on the drink itself: soy milk, what is in it.

Sources

1. ev-soypg-01 · Meta-analysis or systematic review · systematic review and network meta-analysis of randomized controlled trials, GRADE · n = 14 RCTs, 543 participants
   Replacing 500 mL/d of cow's milk with soy drink may reduce low-density lipoprotein (LDL) cholesterol [MD: -0.47 mmol/L (-0.85; -0.10); low certainty], but showed no effect for high-density lipoprotein (HDL) cholesterol [MD: 0.01 mmol/L (-0.03; 0.05); moderate certainty].
   
   Who: adults in randomized trials of at least 3 weeks comparing soy, rice or oat drinks with each other or with cow milk
   Effect: soy drink vs cow milk 500 mL/d: LDL MD -18 mg/dL (95% CI -33 to -3.9; metric: -0.47 mmol/L, -0.85 to -0.10), low certainty; HDL MD 0.01 (-0.03 to 0.05), moderate certainty
   Certainty: Not about pregnant women: no RCT of soy drink in pregnant women was found. This is the best synthesis about the drink itself.
   Adv Nutr, 2026 · checked 2026-10-07 · we read the abstract
2. ev-soypg-02 · Meta-analysis or systematic review · systematic review and meta-analysis of randomized controlled trials · n = 18 studies
   Soy supplementation has no effect on the thyroid hormones and only very modestly raises TSH levels, the clinical significance, if any, of the rise in TSH is unclear.
   
   Who: participants of randomized controlled trials with soy as the intervention and fT3, fT4 and TSH measured
   Effect: fT3 WMD 0.027 pmol/L (95% CI -0.052 to 0.107); fT4 WMD -0.003 pmol/L (-0.018 to 0.011); TSH WMD 0.248 mIU/L (0.001 to 0.494)
   Certainty: Not pregnant; clinical significance of the TSH rise unclear. For a pregnant woman with hypothyroidism: a question for the doctor.
   Sci Rep, 2019 · checked 2026-10-07 · we read the abstract
3. ev-soypg-03 · Meta-analysis or systematic review · systematic review and meta-analysis of randomized trials · n = 47 studies (11 premenopausal)
   In premenopausal women, meta-analysis suggested that soy or isoflavone consumption did not affect primary outcomes estradiol, estrone or SHBG concentrations, but significantly reduced secondary outcomes FSH and LH [by approximately 20% using standardized mean difference (SMD), P = 0.01 and 0.05, respectively].
   
   Who: pre-, peri- and post-menopausal women in randomized or residential crossover trials of soy or isoflavones for 4 or more weeks
   Effect: premenopausal: no effect on estradiol, estrone or SHBG; FSH and LH reduced about 20% (SMD); cycle length +1.05 days (95% CI 0.13 to 1.97)
   Certainty: Not pregnant; shows that the hormonal effect of soy in humans is moderate.
   Hum Reprod Update, 2009 · checked 2026-10-07 · we read the abstract
4. ev-soypg-04 · Observational data · systematic review and meta-analysis of observational studies · n = 19 studies
   Given the high heterogeneity of the included studies, lack of high-quality evidence, and limited studies included for each category, we failed to reach conclusive results regarding the risks of hypospadias, intrauterine growth retardation, maternal anemia, and gestational diabetes mellitus.
   
   Who: pregnant women, vegetarian vs omnivorous diet, observational studies to August 2017
   Effect: low birth weight OR 1.27 (95% CI 0.98 to 1.65); hypospadias, IUGR, anemia, GDM inconclusive
   Certainty: Synthesis of observational studies, hence C. Vegetarianism as a proxy of higher phytoestrogen exposure, not soy milk.
   Crit Rev Food Sci Nutr, 2019 · checked 2026-10-07 · we read the abstract
5. ev-soypg-05 · Observational data · nationwide birth cohort · n = 41578
   Compared with mothers in the reference group (genistein intake 11th-89th percentiles), those in the low intake group (≤10th percentile) had an elevated risk of their sons having hypospadias (multivariable-adjusted OR = 2.8, 95% confidence interval = 1.4-5.8).
   
   Who: mothers who delivered singleton live male births, Japan Environment and Children's Study 2011-2014; FFQ genistein
   Effect: low intake (<=10th percentile) OR 2.8 (95% CI 1.4 to 5.8); high intake (>=90th percentile) OR 0.9 (0.4 to 2.4); 51 cases
   Certainty: 51 cases; observational. Median genistein 15.3 mg/day.
   Urology, 2019 · checked 2026-10-07 · we read the abstract
6. ev-soypg-06 · Observational data · multistate population-based case-control · n = 1250 cases, 3118 controls
   For example, the odds ratio for total phytoestrogen intake was 0.7 (95% confidence interval: 0.5, 1.0).
   
   Who: mothers of hypospadias cases and controls, National Birth Defects Prevention Study, deliveries 1997-2005
   Effect: >=90th vs 11th-89th percentile: OR 0.6 to 0.8; total phytoestrogens OR 0.7 (95% CI 0.5 to 1.0)
   Certainty: The authors ask for verification in other populations.
   Am J Epidemiol, 2013 · checked 2026-10-07 · we read the abstract
7. ev-soypg-07 · Observational data · prospective birth cohort · n = 7928
   Mothers who were vegetarian in pregnancy had an adjusted odds ratio (OR) of 4.99 (95% confidence interval, CI, 2.10-11.88) of giving birth to a boy with hypospadias, compared with omnivores who did not supplement their diet with iron.
   
   Who: boys born to mothers in the Avon Longitudinal Study of Pregnancy and Childhood; prospective questionnaires
   Effect: vegetarian vs omnivore without iron supplements: adjusted OR 4.99 (95% CI 2.10 to 11.88)
   Certainty: Soy was not measured, phytoestrogens are the authors' hypothesis; later JECS and NBDPS did not confirm the signal.
   BJU Int, 2000 · checked 2026-10-07 · we read the abstract
8. ev-soypg-08 · Observational data · prospective birth cohort · n = 84948
   Compared with those in the lowest quintile of soy isoflavone intake, women in the highest quintile were found to have experienced a significantly lower risk of GDM (multivariate relative risk = 0.82; 95% confidence interval: 0.70, 0.95; P for trend = 0.05).
   
   Who: pregnant women, Japan Environment and Children's Study, recruited 2011-2014 at median 12 weeks; FFQ
   Effect: highest vs lowest quintile isoflavone: RR 0.82 (95% CI 0.70 to 0.95); 1904 GDM cases
   Certainty: Protective association with miso and natto; for tofu nonlinear; soy milk not analyzed separately.
   Eur J Nutr, 2021 · checked 2026-10-07 · we read the abstract
9. ev-soypg-09 · Observational data · prospective observational cohort · n = 1495
   Our study found that, compared with the non-consumers, the highest tertile of soy foods intake was associated with a decrease in risk of GDM (RR = 0.73, 95%CI: 0.54-0.99, p = .049).
   
   Who: pregnant women in Southwest China, 3-day 24-h recalls at 6-14 weeks, OGTT at 24-28 weeks
   Effect: highest tertile vs non-consumers: RR 0.73 (95% CI 0.54 to 0.99)
   Certainty: Median soy only 2.9 g/day; observational.
   J Matern Fetal Neonatal Med, 2022 · checked 2026-10-07 · we read the abstract
10. ev-soypg-10 · Observational data · cross-sectional comparison · n = 505
    On the whole, our findings suggest dietary soy consumption during early pregnancy is not associated with the development of thyroid dysfunction or autoimmunity.
    
    Who: women in early pregnancy, Shenyang, China; soy intake frequency by questionnaire
    Effect: no difference in FT4, TSH, TPOAb or thyroid dysfunction across frequent, conventional and occasional soy consumers
    Certainty: Cross-sectional; frequency, not dose.
    J Med Food, 2011 · checked 2026-10-07 · we read the abstract
11. ev-soypg-11 · Randomized controlled trial(s) · randomized clinical trial, 6 weeks · n = 68
    A randomized clinical trial was performed among 68 women with GDM. ... Soy protein consumption in women with GDM significantly improved the glucose homeostasis parameters, triglycerides, and biomarkers of oxidative stress, as well as reductions in the incidence of newborn hyperbilirubinemia and hospitalizations.
    
    Who: women with gestational diabetes, Iran; 0.8 g/kg protein diet, 35% soy protein vs 70% animal protein
    Effect: fasting plasma glucose -12.7 vs +1.4 mg/dL (P < .001); newborn hyperbilirubinemia 8.8% vs 32.4% (P = .01)
    Certainty: A single small RCT; soy protein in the diet, not soy milk; this group of authors had data-integrity concerns in other works: this article is not in Retraction Watch (checked 2026-10-07).
    Journal of Clinical Endocrinology and Metabolism, 2015, 100(12):4654-4661 (Jamilian and Asemi) · checked 2026-10-07 · we read the abstract
12. ev-soypg-12 · Observational data · analysis of a national cross-sectional survey · n = 3976 (intake), 2845 (urinary iodine)
    Exclusive consumers of milk alternatives also had a lower median UIC than cows' milk consumers (79 v. 132 µg/l; P < 0·001) and were classified as iodine deficient by the WHO criterion (median UIC < 100 µg/l), whereas cows' milk consumers were iodine sufficient.
    
    Who: UK National Diet and Nutrition Survey 2014-2017, adults and children; drinks mostly unfortified at the time
    Effect: iodine intake 94 vs 129 mcg/d; median UIC 79 vs 132 mcg/L (P < 0.001)
    Certainty: Not pregnant women separately; pregnant women need more iodine: unfortified soy milk instead of cow's milk removes the main source of iodine.
    Br J Nutr, 2021 · checked 2026-10-07 · we read the abstract
13. ev-soypg-13 · Position of an expert body · government fact sheet · n = —
    The use of soy during pregnancy may be unsafe in amounts greater than those commonly found in foods.
    
    Who: pregnant and breastfeeding women
    Effect: food amounts not flagged; amounts above usual food intake may be unsafe
    Certainty: NCCIH, updated April 2025. That is, a glass of soy milk is a food amount; isoflavone supplements are not.
    NIH National Center for Complementary and Integrative Health, Soy (fact sheet, updated April 2025) · checked 2026-10-07 · we read the section
14. ev-soypg-14 · Position of an expert body · government fact sheet · n = —
    Some people are allergic to soy. In the United States, the presence of soy in packaged foods and dietary supplements must be indicated on the product label.
    
    Who: people with soy allergy
    Effect: soy is a labeled allergen in the US
    Certainty: Axis 6: soy allergy is a direct reason to avoid soy milk, in pregnancy too.
    NIH National Center for Complementary and Integrative Health, Soy (fact sheet, updated April 2025) · checked 2026-10-07 · we read the section