Stevia (steviol glycosides) and blood sugar
What 8 graded rows on stevia (steviol glycosides) agree on for blood sugar, and where they differ. Counted side by side, never averaged.
Consensus finder Rows from the Stevia (steviol glycosides) page, reviewed Oct 21, 2026 How it decides
What the rows agree on
Each block and row is one graded row from the Stevia (steviol glycosides) page, with its PubMed record.
Your question
Stevia (steviol glycosides) and blood sugar
8 rows · 3,003 peopleSplit. 1 row found an improvement, 4 found no difference and 3 had mixed results.
- CGrade C× 3
- DGrade D× 4
- ∅No effect× 1
- Rows
- 8
- Trials
- 52
- People
- 3,003
- Pooled reviews may count some twice
Where they differ
From the rowsThe best-graded row, grade C, is a meta-analysis of 1,439 people (2024) and reported mixed results. 1 row is graded No effect, our grade for adequately powered trials that found no effect.
The 4 rows that found no difference were larger: a median of 35 people, against 12 people for the row that found an improvement.
Only the 4 rows that found no difference name women and men in their population.
The 4 rows that found no difference are newer (2008 to 2026) than the row that found an improvement (2004).
3 rows with mixed results are left out of this comparison.
The rows
1 improved · 4 no difference · 3 mixed- DGrade DGlucose regulation and cardiometabolic markers after 14 days of sweetened drinks · No difference
Crossover RCT · n 39 · 2026 · Healthy, normal-weight adults (Sweden)
220 mg/day · 14 days per drink, with washouts
Glucose regulation did not differ between steviol glycosides, saccharin (220 mg/day) and sucrose (66 g/day). Fasting insulin rose from baseline after stevia, and all three drinks raised after-meal insulin. Triglycerides were higher after sucrose than saccharin; cholesterol, CRP and appetite did not differ.
Row on the Stevia (steviol glycosides) page · PubMed 42514406 (opens PubMed)
- DGrade DGlucose tolerance in type 2 diabetes the day after 3 g · No difference
Crossover RCT · n 30 · 2022 · Adults with type 2 diabetes on metformin or no treatment, University Hospitals Leuven (Belgium), 86.7% men
3 g single dose · Glucose tolerance test 19 hours later
Rebaudioside A was absorbed and converted to steviol and steviol glucuronide, but did not lower the 2-hour glucose curve (−0.7 h·mg/dL, 95% CI −22.3 to 20.9) or change insulin or C-peptide.
Row on the Stevia (steviol glycosides) page · PubMed 36057030 (opens PubMed)
- DGrade DGlucose tolerance and weight with a daily stevia extract · No difference
Open-label trial · n 28 · 2020 · Healthy normal-weight adults, mean age 25
A stevia extract daily (amount not stated in the abstract) · 12 weeks
Glucose and insulin responses to a glucose tolerance test did not change. Weight was stable with stevia (−0.22 kg) while the control group gained 0.89 kg, and energy intake fell in the stevia group.
Row on the Stevia (steviol glycosides) page · PubMed 33036155 (opens PubMed)
- ∅No effectHbA1c and fasting glucose in type 2 diabetes · No difference
RCT · n 122 · 2008 · Men and women aged 33 to 75 with type 2 diabetes
1,000 mg/day · 16 weeks
HbA1c changed by 0.11% with rebaudioside A and 0.09% with placebo (p = 0.355); fasting glucose, insulin, C-peptide, blood pressure, weight and lipids did not differ. Hypoglycaemic episodes were not more common.
Row on the Stevia (steviol glycosides) page · PubMed 18555575 (opens PubMed)
- DGrade DAfter-meal glucose in type 2 diabetes (one dose) · Improved
Crossover RCT · n 12 · 2004 · People with type 2 diabetes
1 g with one meal · Single test meals
Stevioside cut the incremental glucose area under the curve by 18% (P = 0.013) and raised the insulinogenic index by about 40%; insulin, GLP-1 and GIP areas did not change significantly.
Row on the Stevia (steviol glycosides) page · PubMed 14681845 (opens PubMed)
- CGrade CBlood glucose, insulin and HbA1c (pooled) · Mixed
Meta-analysis, 26 studies · n 1,439 · 2024 · Participants of 26 studies
As given in each study; glucose fell at 3,342 mg/day or more in dose-response analysis · Varied by study
Blood glucose fell slightly (−3.84 mg/dL, 95% CI −7.15 to −0.53; low certainty), mainly in people with higher BMI, diabetes or hypertension. Insulin and HbA1c did not change significantly.
Row on the Stevia (steviol glycosides) page · PubMed 39098209 (opens PubMed)
- CGrade CFasting glucose and HbA1c in adults (pooled) · Mixed
Meta-analysis, 12 studies · n 871 · 2024 · Adults in 12 randomized trials (48% women)
As given in each included trial · Varied by trial
Fasting glucose was 4.10 mg/dL lower than control (95% CI −6.55 to −1.65); HbA1c did not differ (0.01%). The overall quality of evidence was rated low.
Row on the Stevia (steviol glycosides) page · PubMed 38512280 (opens PubMed)
- CGrade CDiabetes markers, blood pressure and lipids (pooled) · Mixed
Meta-analysis, 9 studies · n 462 · 2019 · Participants of 9 randomized trials (7 studies)
As given in each included trial · Varied by trial
The review reports a significant fall in systolic pressure (−6.32 mm Hg) and non-significant changes in BMI, diastolic pressure, fasting glucose and cholesterol, with no effect on HbA1c. Heterogeneity was significant.
Row on the Stevia (steviol glycosides) page · PubMed 31438580 (opens PubMed)