Pooling randomized trials, ashwagandha lowered cortisol by 1.16 ug/dL but produced no statistically significant change in Perceived Stress Scale scores.
Anyone weighing cortisol-reduction marketing claims, and readers trying to reconcile conflicting ashwagandha meta-analyses.
P = 0.40The perceived-stress result: ashwagandha lowered cortisol reliably but did not significantly shift how stressed people reported feeling.
What the researchers found
Ashwagandha significantly reduced cortisol by 1.16 ug/dL (95% CI -1.64 to -0.69, P < 0.001) across 7 randomized comparisons.
Ashwagandha did not significantly change perceived stress: SMD = -0.355 (95% CI -1.188 to 0.47, P = 0.40) across 6 randomized comparisons.
The confidence interval for perceived stress crosses zero and is wide, so this is an inconclusive null rather than proof that ashwagandha does nothing subjectively.
This dissociation between a moving biomarker and an unmoved self-report is the paper's central point, and it directly complicates the marketing claim that lowering cortisol makes people feel less stressed.
Why it matters
Nearly every ashwagandha product markets itself on cortisol reduction, with the implicit promise that lower cortisol means feeling calmer. This meta-analysis found that the biomarker moved reliably while the subjective experience did not, at least across the trials pooled here. It is the strongest single argument for treating cortisol change and felt stress relief as separate claims that require separate evidence.
The numbers in context
7 cortisol studies and 6 perceived-stress studies, n = 488 total. Cortisol: -1.16 ug/dL (95% CI -1.64 to -0.69, P < 0.001). Perceived stress: SMD = -0.355 (95% CI -1.188 to 0.47, P = 0.40). Heterogeneity I2 = 50.9%.
How the study worked
PRISMA-guided systematic review and meta-analysis. Four databases (PubMed, CINAHL, Google Scholar, Scopus) were searched for randomized controlled trials published between January 2012 and February 2024. Trials were eligible if they lasted at least two weeks, used oral ashwagandha at 250 mg/day or more, and reported cortisol and Perceived Stress Scale outcomes. Risk of bias was assessed with the Cochrane Risk of Bias 2 tool. Cortisol was pooled as a mean difference and perceived stress as a standardized mean difference.
Who was studied
488 adults across 7 cortisol and 6 perceived-stress randomized comparisons.
What this study cannot tell us
The pooled sample of 488 participants is small, which leaves the perceived-stress analysis underpowered: the wide confidence interval (-1.188 to 0.47) is compatible with anything from a large benefit to a small harm, so the null result should not be read as evidence of no effect. Moderate heterogeneity (I2 = 50.9%) reflects variation in ashwagandha preparation, dose, duration and population across the included trials. Risk of bias was generally moderate, with several trials lacking allocation concealment. Restricting inclusion to trials reporting both cortisol and PSS excludes much of the broader ashwagandha literature, and other meta-analyses using different inclusion criteria have reached a different conclusion about perceived stress. Most included trials were short, so long-term effects remain untested.
How to read the evidence
Meta-analytic synthesis of randomized trials with a registered protocol-style PRISMA method, formal risk-of-bias assessment, and no detected publication bias, which supports a high grade for the cortisol conclusion. The perceived-stress conclusion is graded lower because the pooled sample is small and the confidence interval is wide, making it an inconclusive null rather than a demonstrated absence of effect.
When this study was published
Published 2025, searching literature through February 2024. Current.
The bigger picture
This paper sits in direct tension with Arumugam's 2024 meta-analysis in Explore, which pooled 9 trials in 558 patients and did find a significant Perceived Stress Scale improvement (MD -4.72, 95% CI -8.45 to -0.99) alongside a significant cortisol reduction. The two reviews used different search windows and inclusion rules, which is the most likely source of the disagreement. Taken together they indicate that ashwagandha's effect on cortisol is consistent across reviews while its effect on how stressed people report feeling is not.
Replication
Partially conflicting with Arumugam et al. 2024 (Explore), which found a significant perceived-stress benefit using different inclusion criteria. The cortisol finding is directionally consistent across both reviews.
Funding
No commercial sponsor reported.
Conflicts of interest
No competing interests reported.
Questions still open
- Does the disagreement with Arumugam 2024 come from inclusion criteria, from the choice of mean difference versus standardized mean difference, or from genuine heterogeneity?
- Would a larger pooled sample detect a real but modest perceived-stress benefit?
- Do longer trials, beyond the typically short durations pooled here, close the gap between biomarker and self-report?
- Is a 1.16 ug/dL cortisol reduction clinically meaningful in people without a cortisol disorder?
Common questions
Does this mean ashwagandha does not work for stress?
Why does another meta-analysis disagree?
How big is a 1.16 ug/dL cortisol drop?
Read the original research
Dual impact of Ashwagandha: Significant cortisol reduction but no effects on perceived stress - A systematic review and meta-analysis.
Nutrition and health, 31(4), 1395-1408
Peer-reviewed SAGE journal covering nutrition science and public health nutrition.
Citation
Albalawi, Ahmad Abdualrazag. (2025). Dual impact of Ashwagandha: Significant cortisol reduction but no effects on perceived stress - A systematic review and meta-analysis.. Nutrition and health, 31(4), 1395-1408. https://doi.org/10.1177/02601060251363647