This page has affiliate links to Renue By Science: we earn a commission if you buy through them.
Ashwagandha and sleep: what the randomized trials measured
What randomized trials and one review measured for ashwagandha extract and sleep. Short trials, modest differences. Not a recommendation.
Ashwagandha is a root extract sold for sleep. The fair way to read that claim is to look at randomized, placebo-controlled trials and at what they wrote down: how many people, which extract, how many milligrams, how many weeks, and which sleep numbers moved. This note does that for three trials and for one systematic review that pooled five trials. It is not advice to take the extract.
The pattern in these papers is modest and short. Differences showed up on questionnaires and, in some trials, on wrist actigraphy. The trials ran for six, eight, or ten weeks. The extracts were not the same product. A review of five trials found a small pooled effect, larger in the subgroup of adults with insomnia, at doses of at least 600 mg a day, and at durations of at least eight weeks [4]Citation 4. Small, short, and dependent on which extract was used is the whole summary. The rest of the note is the table behind that sentence.
Grade A. Human RCT or meta-analysis
The trials below are randomized and placebo-controlled, and Cheah and colleagues pooled randomized trials. That is why the grade is A. The grade does not mean the effect is large, and it does not mean the result applies outside the people who were enrolled.
300 mg twice daily for 10 weeks
Langade and colleagues, in 2019, randomized 60 patients with insomnia at one hospital in India, in a 2:1 ratio: 40 to ashwagandha and 20 to placebo. The capsule was a full-spectrum ashwagandha root extract, 300 mg, taken twice a day with milk or water. Placebo capsules contained starch. The planned duration was 10 weeks. Sleep was recorded with Philips Respironics actigraphy (sleep onset latency, total sleep time, sleep efficiency, and wake after sleep onset) and with questionnaires, including the Pittsburgh Sleep Quality Index and the Hamilton Anxiety Rating Scale [1]Citation 1.
Two people, one from each group, did not complete the study. The paper’s per-protocol wording then gives a count that does not add up in the abstract (it says n = 58 and then lists 29 and 19). This note therefore sticks to the figures that are internally consistent: 60 people randomized, two withdrawals, and the 10-week outcome numbers the abstract prints.
Sleep onset latency improved in both the extract group and the placebo group at five and at ten weeks. At ten weeks the extract group was shorter than placebo: 29.00 minutes (standard deviation 7.14) versus 33.94 minutes (standard deviation 7.65), p = 0.019. Sleep efficiency in the extract group moved from 75.63 percent (standard deviation 2.70) at baseline to 83.48 percent (standard deviation 2.83) at ten weeks. In the placebo group it moved from 75.14 percent (standard deviation 3.73) to 79.68 percent (standard deviation 3.59). Sleep quality differed between groups (p = 0.002). The paper also reports change on the anxiety rating scale [1]Citation 1.
Five minutes of sleep-onset latency, and a sleep-efficiency gap of a few points on top of a placebo rise, is a modest difference. Both groups got better, which is what placebo-controlled trials are for: the extract has to beat the improvement that happens anyway. The authors call the extract well tolerated and say larger studies are needed [1]Citation 1. This note does not turn “well tolerated for 10 weeks in this sample” into a general safety certificate.
The anxiety scale is a second questionnaire in a sleep trial. A change on it is not a result about an anxiety disorder, and this note does not treat it as one. It is mentioned because it is in the paper, and then left there.
Eight weeks, two groups, dose not in the abstract
Grade A. Human RCT or meta-analysis
A second randomized, parallel-group trial from Langade and colleagues enrolled 80 adults: 40 in a healthy arm and 40 in an insomnia arm. Within each arm, people were assigned to ashwagandha root extract or placebo for 8 weeks. Outcomes included sleep onset latency, total sleep time, wake after sleep onset, time in bed, sleep efficiency, the Pittsburgh Sleep Quality Index, the Hamilton Anxiety Rating Scale, alertness on waking, and a sleep-quality questionnaire. Adverse events and other medicines were recorded [2]Citation 2.
The abstract reports improvement in sleep measures in both the healthy arm and the insomnia arm, and a larger change in the insomnia arm. In the insomnia data, repeat-measures analysis gave p = 0.013 for sleep onset latency, p < 0.05 for the anxiety scale, p = 0.01 for mental alertness, and p < 0.05 for sleep quality. A two-way analysis pointed to sleep onset latency (p < 0.0001) and sleep efficiency (p < 0.0001) as the measures that moved most, then total sleep time (p < 0.002) and wake after sleep onset (p < 0.040). The authors say the extract was well tolerated across the people in the study, and that further trials are needed before the result can be generalized [2]Citation 2.
The abstract does not state how many milligrams were in the capsule. This note does not look up a blog or a vendor page to fill that hole, and it does not copy a dose from the 2019 trial into this one. They are different studies. A reader who wants the milligram figure needs the full paper, not this summary.
“More change in the insomnia arm than in the healthy arm” is a description of this sample. It is not a rule about who should take the extract. Eighty people, eight weeks, one extract, and an unstated dose in the abstract we are willing to cite, is a narrow base.
120 mg of a different extract for six weeks
Grade A. Human RCT or meta-analysis
Deshpande and colleagues used another preparation. One hundred and fifty healthy adults who scored high on non-restorative sleep measures were randomized, double-blind, to 120 mg of standardized ashwagandha extract (Shoden) once a day, or placebo, for six weeks. One hundred and forty-four people finished. Nobody left because of an adverse event. Sleep quality was the Restorative Sleep Questionnaire, weekly version. Quality of life used the WHOQOL-BREF. Actigraphy recorded sleep latency, efficiency, total sleep time, and wake after sleep onset. Safety labs covered vital signs, blood counts, chemistry, and urine [3]Citation 3.
Self-reported sleep quality rose 72 percent in the extract group and 29 percent in the placebo group (p < 0.001). The placebo rise is large enough that quoting only the 72 percent would mislead. On actigraphy, the extract group differed from placebo after six weeks for sleep efficiency (p < 0.01), total sleep time (p < 0.001), sleep latency (p < 0.01), and wake after sleep onset (p < 0.05). Quality-of-life scores moved in the physical, psychological, and environment domains. The paper reports no treatment-related adverse events [3]Citation 3.
One hundred and twenty milligrams of Shoden is not 300 mg twice a day of the root extract in the 2019 Langade trial, and it is not “ashwagandha” as a single substance. Withanolide content, root versus leaf, and the brand’s standardization are part of the dose. Cheah and colleagues, pooling trials, found the sleep signal more noticeable at doses of at least 600 mg a day [4]Citation 4. A 120 mg trial can still show a difference, as this one did, without matching that subgroup cut-point. The two facts sit side by side: extract identity changes the milligram, and the pooled signal was stronger in the higher-dose subgroup of the review.
The people in Deshpande’s trial were healthy adults with non-restorative sleep scores, not a clinic cohort with a diagnosed sleep disorder. Six weeks is the whole exposure. Actigraphy is motion, not a brain-wave recording. A change in a questionnaire percentage and a change in wrist-monitor minutes are both real outcomes. Neither is a polysomnogram.
What five trials look like together
Grade A. Human RCT or meta-analysis
Cheah and colleagues systematically reviewed randomized trials of ashwagandha extract versus placebo in adults 18 years and older, with sleep quantity and sleep quality as the main outcomes. They analyzed five trials, 400 participants. The pooled effect on overall sleep was a standardized mean difference of −0.59, with a 95 percent confidence interval from −0.75 to −0.42. The I² was 62 percent, which means the trials did not all agree closely. The sleep effect was more pronounced in adults with insomnia, at a daily amount of at least 600 mg, and at a duration of at least eight weeks. The review also reports better mental alertness on rising and a lower anxiety score, and no significant effect on quality of life. No serious side effects were reported. The authors add that data on serious adverse effects are limited, and that more safety data would be needed to judge long-term use [4]Citation 4.
A standardized mean difference of about 0.6 is what statisticians often call a moderate average difference. It is an average across different questionnaires and different extracts. The confidence interval does not cross zero in this analysis, and the heterogeneity is high enough that a single number is a rough guide, not a promise that the next trial will match it. The subgroup pattern, if it holds, says the signal was easier to see when the trial was longer, the dose higher, and the volunteers were people with insomnia. Subgroups inside five trials are fragile. They are a reason to look harder, not a dosing chart.
This note does not re-describe every trial inside that review. Three of the trials discussed above are the ones we read in full enough to report doses and instruments [1]Citation 1[2]Citation 2[3]Citation 3. The review’s other trials are inside the pooled number and are not given a separate write-up here, because this note is limited to the references listed.
Safety, stated only as far as these papers go
Across these reports, the exposure lasted six, eight, or ten weeks [1]Citation 1[2]Citation 2[3]Citation 3. Langade and colleagues described the 10-week and 8-week courses as well tolerated [1]Citation 1[2]Citation 2. Deshpande and colleagues reported no treatment-related adverse events and no dropouts for adverse events, with routine blood and urine tests in the protocol [3]Citation 3. Cheah and colleagues found no serious side effects reported in the five trials, and they said in the same paper that serious-adverse-effect data are limited and that long-term safety is not established by this literature [4]Citation 4.
That is the safety content of these references. They do not, in the texts used here, establish what happens in pregnancy, with thyroid hormone, or with sedative medicines. This note does not add those specifics. The generic point the papers do support is narrower: the trials were short, unwanted effects were not a major feature of the published reports, and the reviewers themselves ask for more safety data before anyone leans on long-term use [4]Citation 4.
None of this is a reason to start or stop a medicine or a supplement. A clinician who knows the person can weigh a six-to-ten-week trial against that person’s own situation. This page cannot.
What was not measured
These trials did not use polysomnography as the reference. They used questionnaires and, in some cases, wrist actigraphy [1]Citation 1[3]Citation 3. Actigraphy is useful and is not a map of sleep stages. A reader who has looked at wearable-versus-laboratory comparisons will know that wrist devices themselves disagree with brain-wave recordings on wake and on stages. An actigraphy change is still a change. It is a change in a proxy.
The samples are single-centre or few-centre trials, largely from India in the Langade and Deshpande papers [1]Citation 1[2]Citation 2[3]Citation 3. Cheah’s pool is 400 people across five trials, not thousands [4]Citation 4. Results can fail to travel when the extract, the questionnaire, or the volunteers change.
No paper in this set followed sleep for a year. No paper here is about children. The outcomes stop at sleep questionnaires, wrist monitors, and the safety checks named above.
Label amount and price
Prices checked 20 Nov 2026. Figures are US store list prices in USD, before shipping. The code HEALTHSPAN is 10 percent off that list price.
LIPO ASHWAGANDHA is 200 mg of root extract per capsule, 90 capsules, $29.95 list ($26.96 with the code), or $0.33 per capsule at list. The page describes it as a full-spectrum root extract with at least 5 percent withanolides. It does not name the extract KSM-66 or Sensoril.
The cited trials did not use this shop product, and their abstracts do not all describe the same extract. Langade 2019 used a full-spectrum root extract, 300 mg twice a day, and the abstract does not name KSM-66 or Sensoril [1]Citation 1. Langade 2021 used a root extract for 8 weeks; the abstract does not give the milligrams and does not name a branded extract [2]Citation 2. Deshpande 2020 used Shoden, 120 mg once a day [3]Citation 3. Cheah’s review found a clearer sleep signal in the subgroup taking at least 600 mg a day [4]Citation 4. Six hundred milligrams of the Renue capsule strength is three capsules, which is the same milligram number as the Langade 2019 daily amount and not the same named extract. A shared milligram is not evidence that the products match.
References
- Langade et al. (2019). Efficacy and safety of ashwagandha (Withania somnifera) root extract in insomnia and anxiety: a double-blind, randomized, placebo-controlled study. Cureus. Back to text
- Langade et al. (2021). Clinical evaluation of the pharmacological impact of ashwagandha root extract on sleep in healthy volunteers and insomnia patients. Journal of Ethnopharmacology. Back to text
- Deshpande et al. (2020). A randomized, double-blind, placebo-controlled study of ashwagandha (Withania somnifera) extract on sleep quality in healthy adults. Sleep Medicine. Back to text
- Cheah et al. (2021). Effect of ashwagandha (Withania somnifera) extract on sleep: a systematic review and meta-analysis. PLoS One. Back to text