Ashwagandha

What it is

Ashwagandha (Withania somnifera) is a medicinal plant of the nightshade family (Solanaceae) that has a long history in Ayurvedic medicine [1] [2]. Trials have tested capsules of concentrated root extract, such as the Shoden extract [3] and a high-concentration full-spectrum root extract [2].

How it might work

Ashwagandha contains active compounds with antioxidant, anti-inflammatory and anti-stress activities [1]. One proposed mechanism is that it moderates the hypothalamus-pituitary-adrenal (HPA) axis, the hormone system that controls cortisol release during stress [3].

What the research says

Anxiety and stress: a 2022 meta-analysis of 12 randomized trials with 1,002 people found that ashwagandha reduced anxiety (standardized mean difference, SMD, -1.55) and stress (SMD -1.75) more than placebo. The certainty of the evidence was rated low, and results varied widely between trials (I² 94% for anxiety) [4]. A 2024 meta-analysis of nine trials (558 people) found lower Perceived Stress Scale scores (mean difference -4.72) and lower Hamilton Anxiety Rating Scale scores (mean difference -2.19) [5]. Another 2024 meta-analysis of five trials (254 people) found lower Hamilton Anxiety scores (mean difference -5.96) [6]. In a 60-day trial of 60 stressed adults, 240 mg a day of an extract reduced Hamilton Anxiety scores more than placebo (P = .040) [3]. A 2026 systematic review of 22 trials reported improvements in stress, depression and anxiety [7]; its effect sizes are much larger than those of the other reviews, and its authors call for better trials to address heterogeneity.

Cortisol: pooled data from nine randomized trials found serum cortisol 2.58 micrograms per decilitre (µg/dL) lower than placebo (95% confidence interval 0.16 to 4.99 lower) [5]. A 2025 meta-analysis of seven trials (488 people) found cortisol lower by 1.16 µg/dL, but no significant change on perceived stress scales [8]. In a 60-day trial of 64 chronically stressed adults, 600 mg a day of root extract (given as 300 mg twice daily) lowered serum cortisol by 27.9% from baseline, against 7.9% with placebo [2]. In a separate 60-day trial of 240 mg a day, morning cortisol fell more than with placebo (P < .001) [3].

Testosterone: a 2026 meta-analysis of 23 randomized trials (1,706 adults) found testosterone about 57 ng per decilitre (ng/dL) higher than placebo in men, with no significant change in women; the authors flagged heterogeneity and limited data and called for better-standardized trials [9]. Small trials agree: in 57 young men doing resistance training, 600 mg a day raised testosterone by 96 ng/dL over 8 weeks, against 18 ng/dL with placebo [10]; in aging, overweight men, salivary testosterone rose 14.7% more than with placebo [11]; and in stressed, overweight adults, free testosterone rose significantly in the men taking ashwagandha [12].

Sleep: a 2021 meta-analysis of five trials (400 people) found a small but significant improvement in overall sleep (SMD -0.59). The effect was larger in adults with insomnia, at doses of 600 mg a day or more, and with courses of eight weeks or longer [13]. A 2024 meta-analysis of five trials found significant effects on sleep onset, total sleep time, sleep efficiency and Pittsburgh Sleep Quality Index scores, but not on wake time after sleep onset or time in bed [6].

Doses used in studies

In the trials cited here, one used 240 mg once a day for 60 days [3], and another used 600 mg a day, given as 300 mg twice daily, for 60 days [2]. A thyroid pilot trial used 600 mg a day for eight weeks [14]. Across the reviews, trial durations ran from 30 to 112 days [1]. The sleep review found larger effects at 600 mg a day or more for eight weeks or longer [13], and the stress analysis favoured 300 to 600 mg a day [4]. The testosterone trials used 600 mg a day for eight weeks [10], 21 mg a day of withanolide glycosides [11], and 400 mg a day for 12 weeks [12].

Safety and interactions

• Side effects: in a placebo-controlled trial, side effects were mild and comparable between groups, and no serious adverse events were reported [2]. A 2024 review reported mild to moderate adverse events in four of its trials [5]. In a thyroid trial, 4 of 50 people had mild, temporary side effects, one on ashwagandha and three on placebo [14]. The abstracts do not name the specific side effects [2].
• Upper intake limits: the studies cited on this page set no upper intake limit or toxic dose [4] [5] [13]. A 2023 review says the long-term effects are not yet fully understood [1], and a 2021 sleep review says more safety data are needed before long-term use [13].
• Liver injury: case series link ashwagandha to liver injury. In five people from Iceland and the US, all developed jaundice and itching 2 to 12 weeks after starting; the jaundice and itching lasted 5 to 20 weeks in some, and liver tests returned to normal within 1 to 5 months in four of them. None developed liver failure, and in one case rhodiola, another supplement, may also have contributed [15]. In a series of eight people from India, five had chronic liver disease, and three developed acute-on-chronic liver failure and died [16]. People with existing liver disease were the group at highest risk in that series [16].
• Thyroid: in a small placebo-controlled pilot trial of 50 adults with subclinical hypothyroidism (raised TSH), 600 mg a day for eight weeks improved TSH, T3 and T4 levels and normalized thyroid indices compared with placebo [14]. Because the extract changes thyroid hormone levels, people with thyroid conditions or on thyroid medicines are a group to ask a doctor about first. The trial does not report results in people with an overactive thyroid [14].
• Medicine interactions: the studies cited on this page do not report interactions with medicines, including sedatives, benzodiazepines, immunosuppressants, blood pressure or blood sugar medicines [3] [5] [13]. Interactions are therefore not established by these sources.
• Pregnancy and breastfeeding: the studies cited on this page report no data on pregnancy or breastfeeding [2] [3], so safety in these groups is not established by these sources.
• Kidney disease and bleeding disorders: the studies cited on this page report no results for people with kidney disease or bleeding disorders [2] [3].

Caveats and open questions

Trials are few and small. The 2022 review pooled 12 trials with 1,002 people, and its anxiety results varied widely (I² 94%) [4]. Some trials were run at single centres, including in India [2] [14]. One author of a 2019 trial had received past funding from an extract maker, although that trial was managed independently [3].

Long-term effects are not well understood. None of the studies in a 2023 systematic review assessed the impact of cortisol reduction on adrenal function or on long-term effects [1]. Larger trials with more diverse populations and doses are needed [3].

Cortisol and perceived stress can move differently: the 2025 review found lower cortisol without a significant change in perceived stress [8].