Is stevia safe during pregnancy?

Nobody has tested it in pregnant women, and no study we found reports stevia on its own. The WHO review behind its 2023 sweetener guideline found no randomized trial of any sweetener in pregnancy. The best summary is a 2024 meta-analysis of 19 studies with 203,706 participants, mostly cohorts. It linked low-calorie sweeteners as a class with an 11% higher risk of preterm birth (RR 1.11, 95% CI 1.07 to 1.16), and it did not separate stevia from the rest. EFSA's safety review of steviol glycosides themselves found no link to harm to reproduction or the developing child, and set an acceptable daily intake of 4 mg per kg of body weight.

Unsettled. The answer is unclear for a plain reason. The human signal comes from cohorts of women drinking diet soft drinks, and the Danish data were collected in 1996 to 2002, before the EU approved steviol glycosides in 2011. The reassuring side is toxicology in animals. Neither half measures stevia in pregnant women, so neither half can settle the question.

What the trials found

There are no trials. WHO's systematic review identified 50 randomized trials of non-sugar sweeteners in adults and children, and none in pregnant women. Its pregnancy evidence came from cohort studies. Our own searches of Europe PMC and ClinicalTrials.gov found no trial of stevia in pregnancy either.

Preterm birth

The 2024 meta-analysis in Clinical Nutrition ESPEN pooled 19 studies of women exposed to low-calorie sweeteners before conception or during pregnancy. Preterm birth was 11% more likely, with an interval of 1.07 to 1.16. The authors read this as an association and asked for randomized trials. The largest single cohorts point the same way. In 59,334 Danish women, one or more artificially sweetened soft drinks a day went with 38% higher odds of preterm delivery (OR 1.38, 95% CI 1.15 to 1.65), in data gathered before stevia was approved in the EU. In 60,761 Norwegian women, more than one diet drink a day went with 11% higher odds (OR 1.11, 95% CI 1.00 to 1.24). In the same cohort, sugary drinks carried 25% higher odds (OR 1.25, 95% CI 1.08 to 1.45), which suggests that something shared by heavy drinkers of both kinds, not the sweetener alone, may explain part of the link.

Gestational diabetes

The same 2024 meta-analysis put gestational diabetes 42% higher among sweetener users, with an interval from 0.98 to 2.04. The authors call that estimate imprecise, because it runs from slightly lower risk to double. In a Chilean cohort of 1,472 pregnant women, 41% of whom ate steviol glycosides, any non-nutritive sweetener went with 58% higher odds of gestational diabetes (OR 1.58, 95% CI 1.10 to 2.26). The signal the authors named was sucralose, and the abstract names no link for steviol glycosides.

The baby's weight and later weight

A 2025 meta-analysis of 17 studies, 14 of them cohorts, found sweetener use in pregnancy was not associated with birth weight across four cohorts (SMD 0.03, 95% CI -0.03 to 0.08). Overweight at one year did not reach significance either (OR 1.19, 95% CI 0.81 to 1.58, two cohorts). The evidence for every outcome was rated low or very low. Two single cohorts found more. In 3,033 Canadian mother and infant pairs, daily diet drinks in pregnancy went with about twice the odds of infant overweight at one year (OR 2.19, 95% CI 1.23 to 3.88), and only 5.1% of the women drank them daily. In the Danish cohort, children of mothers who drank at least one a day had 26% higher odds of overweight at 18 (OR 1.26, 95% CI 1.12 to 1.42). Neither cohort named the sweetener, and the Danish pregnancies predate stevia in the EU.

Who should be careful

Not for everyone. Anyone using a lot of it. EFSA set the acceptable daily intake for steviol glycosides at 4 mg per kg of body weight, as steviol equivalents, and warned that adults and children could exceed it if the sweeteners were used at the maximum levels proposed. The one survey of intake in pregnancy we have found far lower use. Among 601 pregnant women in Chile, 78% used stevia, the average intake was 12% of that limit, and no woman went above it.

Leaf and crude extracts are a different product. In the US only high-purity steviol glycosides, at least 95% pure, are generally recognized as safe. Stevia leaf and crude stevia extracts are not, and they cannot be imported as sweeteners. The safety reviews below are about the purified sweetener.

Breastfeeding has almost no data. In 682 US mother and infant pairs, infants exposed to low-calorie sweeteners through breast milk had 2.78 times the odds of vomiting (95% CI 1.05 to 7.34), with no link to infant weight. That was a cross-sectional analysis of diet drinks in general, and it did not look at stevia. We found no study measuring stevia in human breast milk, and LactMed, the US database of drugs and chemicals in milk, has no record for stevia. That is a gap in the data, not a sign of safety.

The animal evidence is narrow. Female rats given rebaudioside A at the acceptable daily intake, from four weeks before mating until weaning, showed no change in glucose tolerance or heart function. Their offspring were not studied, and the authors say that study is still needed.

What expert bodies say

WHO's 2023 guideline advises against non-sugar sweeteners for weight control, names stevia and stevia derivatives among the sweeteners it covers, and says the recommendation applies to pregnant and breastfeeding women, with the exception of people with diabetes. It is a conditional recommendation about weight control, and WHO rated the pregnancy evidence low certainty. Pooling three cohorts of 129,009 pregnant women, WHO found higher sweetener intake associated with 25% higher odds of preterm birth (OR 1.25, 95% CI 1.07 to 1.46).

EFSA found steviol glycosides were not genotoxic or carcinogenic and were not linked to adverse effects on the reproductive system or the developing child. The FDA cites the same acceptable daily intake of 4 mg per kg, set by JECFA. In 2012 the Academy of Nutrition and Dietetics listed stevia among seven non-nutritive sweeteners approved in the US and said all of them were safe. None of these bodies tested stevia in pregnant women. More on stevia itself is on the stevia page.

How we searched

What would change this answer

• A cohort that reports steviol glycoside intake separately and follows it to preterm birth or gestational diabetes. No study we found does this.
• A randomized trial in pregnant women. WHO found none, and none was registered on ClinicalTrials.gov on 7 October 2026.
• A measurement of steviol in human breast milk.
• A rodent study that follows the offspring of mothers given rebaudioside A, which the authors of the rat study call mandatory.

Sources

1. ev-stpg-01 · Observational data · systematic review and meta-analysis of clinical trials, cohort and case-control studies, GRADE · n = 203,706
   LCS consumption during pregnancy was associated with 11% increased risk of preterm birth (RR = 1.11, 95% CI: 1.07-1.16, I2 = 0.01%) and 42% increased risk of gestational diabetes (RR = 1.42, 95% CI: 0.98-2.04, I2 = 67.60%) compared with no consumption, however, the effect size for gestational diabetes was not precise as the 95% CI indicated that the effect estimate could range from 2% lower risk to 204% (or 2.04 times) higher risk.
   
   Who: Women exposed to low-calorie sweeteners before conception or during pregnancy, 19 studies
   Effect: preterm birth RR 1.11 (95% CI 1.07 to 1.16); gestational diabetes RR 1.42 (95% CI 0.98 to 2.04); no difference in gestational weight gain
   Certainty: Synthesis mostly of cohorts: effect estimates from observational data, causality not shown (the authors call for RCTs). Sweeteners as a class, not stevia separately; steviol glycosides not separated in the estimates. No. 6 2026-10-07: level A → C (§2: A only for a synthesis of RCTs; pooled estimates from cohorts).
   Clinical Nutrition ESPEN, 2024 · checked 2026-10-07 · we read the abstract
2. ev-stpg-02 · Observational data · systematic review and meta-analysis of trials and cohort studies (14 cohorts, 1 RCT), GRADE · n = 17 studies
   Any LCS consumption during pregnancy compared with no consumption was not associated with birth weight [SMD (95% CI) = 0.03 (-0.03, 0.08)] (four cohort studies).
   
   Who: Children of women who used low-calorie sweeteners before or during pregnancy or while breastfeeding, 17 studies
   Effect: birth weight SMD 0.03 (95% CI -0.03 to 0.08), 4 cohorts; BMI z-score at birth to 7 years not different; overweight at 1 year OR 1.19 (0.81 to 1.58), 2 cohorts
   Certainty: Full text (PMC11650048, Europe PMC): 14 cohorts and one RCT (xylitol in gum, not stevia); the evidence quality of all outcomes is GRADE low or very low. Frequent intake (≥1 serving/day) gave SMD 0.04 (0.00-0.08) for birth weight. No. 6 2026-10-07: level A → C (§2: estimates from cohorts, only one RCT and not about stevia).
   Maternal & Child Nutrition, 2025 · checked 2026-10-07 · we read the fulltext
3. ev-stpg-03 · Position of an expert body · WHO guideline built on a systematic review of RCTs and observational studies · n = —
   No RCTs in pregnant women were identified.
   
   Who: Systematic review behind the WHO non-sugar sweetener guideline, adults, children and pregnant women
   Effect: 0 RCTs in pregnant women; pregnancy evidence from 17 cohort studies (12 cohorts)
   Certainty: Axis 9 as a fact: there are no randomized data on sweeteners in pregnant women. Our search of Europe PMC and ctgov_search also found nothing for stevia.
   WHO, 2023, Use of non-sugar sweeteners: WHO guideline · checked 2026-10-07 · we read the section
4. ev-stpg-04 · Position of an expert body · WHO guideline evidence summary of a meta-analysis of prospective cohort studies · n = 129,009
   Higher intakes of NSS were associated with a 25% increase in risk of preterm birth, as assessed by meta-analyses of three prospective cohort studies (odds ratio [OR] 1.25; 95% CI: 1.07 to 1.46; 129 009 pregnant women).
   
   Who: 129,009 pregnant women in three prospective cohort studies
   Effect: preterm birth OR 1.25 (95% CI 1.07 to 1.46); low certainty; mainly late preterm
   Certainty: The number from T2 rests on the primary cohorts ev-stpg-07 (Denmark) and ev-stpg-08 (Norway) and on the meta-analysis ev-stpg-01. WHO itself writes about possible reverse causation and residual confounding; sweeteners in drinks, not stevia separately.
   WHO, 2023, Use of non-sugar sweeteners: WHO guideline · checked 2026-10-07 · we read the section
5. ev-stpg-05 · Position of an expert body · WHO guideline remark · n = —
   With the exception of individuals with diabetes (as noted below), this recommendation is relevant for everyone: children and adults of any age, including pregnant and lactating women.
   
   Who: General population including pregnant and lactating women, WHO 2023 guideline
   Effect: conditional recommendation, no effect size
   Certainty: A conditional recommendation for weight control, not a statement about toxicity. For pregnant women WHO rated the certainty of evidence as low.
   WHO, 2023, Use of non-sugar sweeteners: WHO guideline · checked 2026-10-07 · we read the section
6. ev-stpg-06 · Position of an expert body · WHO news release · n = —
   Common NSS include acesulfame K, aspartame, advantame, cyclamates, neotame, saccharin, sucralose, stevia and stevia derivatives.
   
   Who: Scope of the WHO 2023 non-sugar sweetener guideline
   Effect: stevia included in scope, no effect size
   Certainty: Scope of the recommendation: stevia as a "natural" sweetener is not excluded.
   WHO, 2023, news: WHO advises not to use non-sugar sweeteners for weight control in newly released guideline · checked 2026-10-07 · we read the section
7. ev-stpg-07 · Observational data · prospective cohort · n = 59,334
   In comparison with women with no intake of artificially sweetened carbonated soft drinks, the adjusted odds ratio for women who consumed ge 1 serving of artificially sweetened carbonated soft drinks/d was 1.38 (95% CI: 1.15, 1.65).
   
   Who: 59,334 pregnant women in the Danish National Birth Cohort, 1996-2002
   Effect: adjusted OR 1.38 (95% CI 1.15 to 1.65) for at least 1 serving/day; OR 1.78 (1.19 to 2.66) for at least 4 servings/day; no association for sugar-sweetened drinks
   Certainty: Observational; data 1996-2002, before steviol glycosides were permitted in the EU (2011), so about other sweeteners.
   Am J Clin Nutr, 2010 · checked 2026-10-07 · we read the abstract
8. ev-stpg-08 · Observational data · prospective cohort · n = 60,761
   A high intake of AS beverages was associated with preterm delivery; the adjusted OR for those drinking >1 serving/d was 1.11 (95% CI: 1.00, 1.24). Drinking >1 serving of SS beverages per day was also associated with an increased risk of preterm delivery (adjusted OR: 1.25; 95% CI: 1.08, 1.45).
   
   Who: 60,761 pregnant women in the Norwegian Mother and Child Cohort Study
   Effect: artificially sweetened >1 serving/day adjusted OR 1.11 (95% CI 1.00 to 1.24); sugar-sweetened >1 serving/day adjusted OR 1.25 (1.08 to 1.45)
   Certainty: Replication of the Danish signal: the effect is weaker and the same for sugary drinks, which points to lifestyle confounding.
   Am J Clin Nutr, 2012 · checked 2026-10-07 · we read the abstract
9. ev-stpg-09 · Observational data · prospective birth cohort · n = 3,033
   Compared with no consumption, daily consumption of artificially sweetened beverages was associated with a 0.20-unit increase in infant BMI z score (adjusted 95% CI, 0.02-0.38) and a 2-fold higher risk of infant overweight at 1 year of age (adjusted odds ratio, 2.19; 95% CI, 1.23-3.88).
   
   Who: 3,033 mother-infant dyads, Canadian Healthy Infant Longitudinal Development Study
   Effect: infant BMI z score +0.20 (95% CI 0.02 to 0.38); overweight at 1 year adjusted OR 2.19 (1.23 to 3.88)
   Certainty: Observational; daily intake in only 5.1% of women. Sweetener type not separated.
   JAMA Pediatr, 2016 · checked 2026-10-07 · we read the abstract
10. ev-stpg-10 · Observational data · secondary analysis of a prospective cohort · n = 1,472
    The adjusted model showed a significant association between the consumption of any NNS and sucralose and the risk of GDM (OR for any NNS = 1.58; 95% CI: 1.10-2.26; P = 0.014; OR sucralose = 1.44; 95% CI 1.06-1.95; P = 0.020).
    
    Who: 1,472 pregnant women in the CHiMINCs-II cohort, Santiago, Chile
    Effect: any NNS adjusted OR 1.58 (95% CI 1.10 to 2.26); sucralose OR 1.44 (1.06 to 1.95); steviol glycosides consumed by 41.1%, no significant association reported
    Certainty: Observational, 24-hour recall. For steviol glycosides no significant association is named in the abstract (same in cards_daily_stevia).
    Nutrition, 2024 · checked 2026-10-07 · we read the abstract
11. ev-stpg-11 · Position of an expert body · agency safety assessment news · n = —
    Toxicological testing showed that the substances are not genotoxic, nor carcinogenic, or linked to any adverse effects on the reproductive human system or for the developing child. ... The Panel points out, however, that this ADI could be exceeded by both adults and children if these sweeteners are used at the maximum levels proposed by the applicants.
    
    Who: EFSA ANS Panel safety assessment of steviol glycosides (E 960), 2010
    Effect: ADI 4 mg/kg bw/day (as steviol equivalents); no reproductive or developmental toxicity
    Certainty: Toxicological assessment (animals), not a study of pregnant women. For a 143 lb (65 kg) woman the ADI = 260 mg steviol equivalents per day.
    EFSA, 2010, news: EFSA evaluates the safety of steviol glycosides · checked 2026-10-07 · we read the section
12. ev-stpg-12 · Laboratory or animal data · animal study · n = —
    Female rats were exposed to RebA (4 mg steviol equivalents/kg body weight/day) in the drinking water from 4 weeks before mating until weaning. ... This work suggests that this RebA dose is safe for female consumption during the reproductive stage, from a cardiometabolic perspective. However, studies on the effects of RebA exposure on the offspring are mandatory.
    
    Who: Female rats, rebaudioside A 4 mg steviol equivalents/kg bw/day in drinking water
    Effect: no change in glucose tolerance or cardiac function; lower fasting glycemia and cholesterol; smaller cardiomyocytes
    Certainty: Animals; offspring were not assessed, the authors say this is mandatory. Axis 9: there are no human data on stevia in pregnant women.
    Biology (Basel), 2024 · checked 2026-10-07 · we read the abstract
13. ev-stpg-13 · Observational data · cross-sectional analysis of a longitudinal study · n = 682
    While no significant associations were observed between infants' LCS exposure and BMI-for-age or risk of overweight, infants' LCS exposure was associated with a 2.78-fold increased risk of vomiting (95% confidence interval 1.05-7.34).
    
    Who: 682 mother-infant dyads at 3 months postpartum, Infant Feeding Practices Study II, 2005-2007
    Effect: vomiting OR 2.78 (95% CI 1.05 to 7.34); no association with BMI-for-age or overweight
    Certainty: Cross-sectional, small sample, exposure is diet drinks overall, not stevia. There are no data on stevia in breast milk (LactMed has no entry).
    Nutrients, 2021 · checked 2026-10-07 · we read the abstract
14. ev-stpg-14 · Observational data · longitudinal birth cohort · n = 72,821
    We found increased odds of overweight in 7-, 11-, 14- and 18-year-old offspring whose mothers reported drinking ≥ 1 artificially sweetened beverage daily during pregnancy compared with no consumption (18 years: adjusted OR 1·26 (95 % CI 1·12, 1·42)).
    
    Who: 72,821 women with a pregnancy food frequency questionnaire and their offspring followed to 18 years, Danish National Birth Cohort
    Effect: overweight at 18 years adjusted OR 1.26 (95% CI 1.12 to 1.42); raised odds also at 7, 11 and 14 years
    Certainty: Observational; pregnancies 1996-2002, before stevia was permitted in the EU, so about other sweeteners. For sugary drinks the odds were lower, which points to residual confounding.
    Br J Nutr, 2025 · checked 2026-10-07 · we read the abstract
15. stv1-daily-r1-13 · Position of an expert body · agency guidance page · n = —
    Rebaudioside A, a type of steviol glycoside, is 12 mg/kg bw/d. This is based on the ADI for steviol glycosides: 4 mg/kg bw/d, expressed as steviol equivalents, set by the Joint FAO/WHO Expert Committee on Food Additives, on which the FDA participates.
    
    Who: US regulatory safety assessment of high-purity steviol glycosides
    Effect: ADI 4 mg/kg bw/day as steviol equivalents; 12 mg/kg bw/day for rebaudioside A
    Certainty: Authority position; ADI is a safety threshold, not an effect.
    U.S. Food and Drug Administration, Aspartame and Other Sweeteners in Food · checked 2026-09-29 · we read the section
16. stv1-daily-r1-14 · Position of an expert body · agency guidance page · n = —
    The use of stevia leaf and crude stevia extracts is not considered GRAS, and their import into the U.S. is not permitted for use as sweeteners.
    
    Who: US regulatory status of stevia-derived sweeteners
    Effect: high purity (95% minimum) steviol glycosides GRAS; whole leaf and crude extracts not GRAS
    Certainty: US regulatory status.
    U.S. Food and Drug Administration, Aspartame and Other Sweeteners in Food · checked 2026-09-29 · we read the section
17. stv1-daily-r1-15 · Observational data · cross-sectional survey · n = 601
    The most consumed sweetener was sucralose (95.6%), followed by acesulfame k (80.6%), stevia (78.3%), aspartame (75.1%), saccharin (14.8%), and cyclamate (10%). ... However, except stevia, which reached 12%, average values were under 5% of the acceptable daily intake. No pregnant woman in the sample exceeded the acceptable daily intake
    
    Who: 601 pregnant women in the two main cities of Chile
    Effect: stevia used by 78.3%; mean intake 12% of ADI; no woman above ADI
    Certainty: Survey consumption data.
    Nutr Hosp, 2019 · checked 2026-09-29 · we read the abstract
18. stv1-daily-r1-17 · Position of an expert body · professional position statement · n = —
    Seven NNS are approved for use in the United States: acesulfame K, aspartame, luo han guo fruit extract, neotame, saccharin, stevia, and sucralose. ... All NNS approved for use in the United States are determined to be safe.
    
    Who: US dietetic professional position paper, 2012
    Effect: position statement, no effect size
    Certainty: Association position, 2012.
    J Acad Nutr Diet, 2012 · checked 2026-09-29 · we read the abstract