Ashwagandha Dosage, Onset and Timing, Read Off the Trials

An ashwagandha dose is not a number on its own. Published trials used between 125 mg and 600 mg a day, but each of those figures belongs to a specific named extract: KSM-66 at 300 mg twice daily, Sensoril at 125 mg, 250 mg or 500 mg once at night, Shoden at 240 mg once after dinner. The same milligram figure can carry more than ten times the withanolide content depending on which of those you hold. And across those trials the first measured difference came at week 4 or later, never on day two.

That is the problem with the ashwagandha dosage question as it is usually answered. “Take 300 mg to 600 mg daily” gives you nothing to act on, because it names no extract, no plant part, no standardisation and no measurement point. Below is the dose, the extract and the schedule for every trial behind the numbers everyone repeats, then what is known about onset, duration, morning against night, course length and stopping. Ashwagandha evidence and safety in the round belong to a separate article.

How to read a dose on an ashwagandha label

Six things decide whether a milligram figure means anything. Check them in this order, on the label, before you compare two products or two studies.

The plant part. Root only, or root and leaf. KSM-66 is a root-only extract. Sensoril is drawn from root and leaf. Withanolide profiles differ between the two, so a leaf-inclusive extract and a root-only extract are different materials at the same weight.

Extract or powder. A raw root powder and a concentrated extract are separated by the extraction ratio. Traditional preparations use grams of powder. Modern trials use hundreds of milligrams of extract. A label that says “ashwagandha 1,000 mg” without saying which of the two it is has answered nothing.

The named extract. KSM-66, Sensoril and Shoden are the three trade names that carry most of the clinical literature. If the label names one, you can find the trials that used it. If it says only “ashwagandha extract”, you cannot, and the published doses below do not transfer to it.

The standardisation. This is the figure that matters most and the one most often missing. KSM-66 is standardised to at least 5 per cent withanolides by HPLC. Shoden is standardised to 35 per cent withanolide glycosides. Those are not the same measurement, and the numbers are not interchangeable.

Which withanolides. “Withanolides” and “withanolide glycosides” are different analytes. In the Nutrients trial of Sensoril, nine withanolides and one flavonoid glycoside were quantified by UPLC. Two labels can both say 5 per cent and be measuring different things by different methods.

The daily total against the per capsule figure. Trials report daily totals. Labels report per capsule. A trial giving 300 mg twice daily is a 600 mg daily total in two servings, which is not the same schedule as 600 mg once.

If a product fails four of these six, its milligram figure is decoration. That is the position most of the shelf is in.

The three standardised extracts the trials used

Three named extracts account for most of the ashwagandha dosage figures in circulation. They are not variants of one dose. They are different materials, standardised on different analytes, tested at different amounts.

KSM-66, made by Ixoreal Biomed, is a root-only, water-based extract standardised to at least 5 per cent withanolides by HPLC. It is the extract in Chandrasekhar and colleagues (2012), in Salve and colleagues (2019), and in the Langade insomnia study (2019). Its trials cluster on 300 mg twice daily.

Sensoril, now held by Kerry, is a water-extracted root and leaf preparation. In the 2024 Nutrients trial its withanolide constituents were quantified at 0.4 to 1 per cent each across nine compounds. Its trials use much smaller daily amounts, 125 mg to 500 mg, in a single serving.

Shoden, made by Arjuna Natural, is standardised to 35 per cent withanolide glycosides and delivered in a bead format. Its trials use the smallest weights of all: 240 mg once daily in the stress study, 120 mg once daily in the sleep study, and in one crossover study the dose was reported not as extract weight but as 21 mg of withanolide glycosides a day.

Extract	Plant part	Standardised to	Doses used in trials	Typical schedule
KSM-66	Root only	5 per cent withanolides or more, HPLC	250 mg, 600 mg daily	Two servings, after food
Sensoril	Root and leaf	Nine withanolides quantified by UPLC	125 mg, 250 mg, 500 mg daily	One serving at night
Shoden	Root and leaf beads	35 per cent withanolide glycosides	120 mg, 240 mg daily	One serving after dinner

Compare the standardisation first and the milligrams second. A 240 mg Shoden serving and a 600 mg KSM-66 serving are both credible trial doses, and neither is the bigger dose in any meaningful sense.

What ashwagandha dosage each trial actually used

Here are the numbers, with the population, the duration and what was measured. These are what the trials used. They are not a recommended dosage for you, and none of them was tested against a condition you can self-diagnose.

Chandrasekhar, Kapoor and Anishetty (2012), in the Indian Journal of Psychological Medicine, gave 64 adults with chronic stress, aged 18 to 54, one 300 mg KSM-66 capsule twice daily for 60 days. Sixty-one completed. Perceived stress scores and serum cortisol were lower in the extract arm than in placebo. The authors themselves noted that the sample was not large and asked for longer studies in more diverse populations.

Salve, Pate, Debnath and Langade (2019), in Cureus, ran three arms in 60 stressed but healthy adults aged 18 to 55: KSM-66 at 250 mg a day, KSM-66 at 600 mg a day, and placebo, for 8 weeks, taken as one capsule twice a day after food with milk or water. Mean Perceived Stress Scale scores fell from 22.95 to 18.85 at week 4 and 14.15 at week 8 on 600 mg, against 22.70 to 19.89 and 16.63 on placebo. Serum cortisol went from 16.12 to 10.86 micrograms per decilitre on 600 mg, against essentially no change on placebo. The extract was supplied by its manufacturer, which is worth knowing when you read the effect sizes.

Pandit and colleagues (2024), in Nutrients, tested Sensoril at 125 mg, 250 mg and 500 mg a day against placebo for 8 weeks in adults aged 18 to 60 with stress lasting more than three months. One hundred and thirty one enrolled, 98 completed per protocol. The study product was “a single capsule at night, approximately half an hour before bedtime”. The Perceived Stress Scale first separated from placebo at week 4, and the 500 mg arm ended lowest. The study was funded by the extract’s owner and lost a large number of participants to the pandemic.

Lopresti, Smith, Malvi and Kodgule (2019), in Medicine, gave 60 adults aged 18 to 65 with mild stress either 240 mg of Shoden or placebo once daily after dinner with 250 ml of water, for 60 days. Anxiety scale scores and morning cortisol were lower in the extract arm. The authors listed the small sample, the healthy population and the variety of extracts across the literature as reasons dose comparisons across studies are unreliable.

Langade and colleagues (2019), in Cureus, gave 60 adults with diagnosed insomnia 300 mg of KSM-66 twice daily with milk or water for 10 weeks, with assessments at baseline, week 5 and week 10. Sleep onset latency at week 10 was 29.00 minutes in the extract arm against 33.94 minutes in placebo. That is a difference of about five minutes on a measured average, which is a more modest number than the headlines built on it.

Marchi and colleagues (2025), in BJPsych Open, pooled 14 randomised trials covering 713 people. The median dose across them was 600 mg a day and the median follow-up was 8 weeks. They flagged substantial heterogeneity between studies, problems with allocation and blinding, and incomplete reporting of extract composition. That last point is the reason this page starts with the label rather than the milligrams.

Trial	Extract	Daily dose	Duration	Population
Chandrasekhar 2012	KSM-66	600 mg, split	60 days	64 adults, chronic stress
Salve 2019	KSM-66	250 mg or 600 mg	8 weeks	60 stressed adults
Langade 2019	KSM-66	600 mg, split	10 weeks	60 adults with insomnia
Lopresti 2019	Shoden	240 mg	60 days	60 adults, mild stress
Pandit 2024	Sensoril	125, 250 or 500 mg	8 weeks	131 chronically stressed adults

The pattern is narrow: 8 weeks, small samples, healthy or mildly stressed adults, manufacturer funding in most cases. Take that as the boundary of what any dose figure is entitled to say.

How long ashwagandha takes to work

Nobody has measured a change on day two, because no trial looked. This is the most important thing on the page and it is missing from almost every article that answers the question.

The earliest scheduled assessment in the trials above is week 2, in the Nutrients Sensoril study. The first point at which a measurement separated from placebo in that study was week 4. In Salve 2019 the assessment points were baseline, week 4 and week 8, and both the 250 mg and 600 mg arms had moved by week 4 with a larger gap by week 8. Langade 2019 measured at week 5 and week 10. Chandrasekhar 2012 ran 60 days.

So the evidence base has a shape, and the shape is weeks. A meta-analysis of 14 trials found a median follow-up of 8 weeks. If you are trying a standardised extract, the observation window that matches the literature is 8 weeks of consistent daily servings, assessed once at the start and once at the end. Judging it after three days is judging something no trial has ever measured.

What to do with that: write down the date you start, and do not form a view before week 4. The same discipline applies to shilajit onset, for the same reason.

How long a single serving stays in the blood

Onset over weeks and duration within a day are two different questions, and the second one has an actual answer from pharmacokinetics.

Kim, Venkatesan, Rathi and Antony (2023), in Heliyon, ran a double-blind crossover in 16 healthy adults, 15 completing, comparing two extracts at an equal 185 mg of total withanolides under fasting conditions with a 7 day washout. The extract standardised to 35 per cent withanolide glycosides reached a peak plasma concentration of 60.66 ng/ml at 2.277 hours, with a half life of 9.756 hours and a mean residence time of 14.458 hours. The extract standardised to 2.5 per cent withanolides peaked at 10.789 ng/ml at 1.5 hours, with a half life of 1.882 hours and a mean residence time of 3.503 hours.

Read those two rows next to each other. At the same withanolide load, one extract stays measurable in plasma roughly five times longer than the other. Duration is a property of the extract, not of ashwagandha in general.

Measure	35 per cent glycoside extract	2.5 per cent withanolide extract
Time to peak	2.277 hours	1.5 hours
Peak concentration	60.66 ng/ml	10.789 ng/ml
Half life	9.756 hours	1.882 hours
Mean residence time	14.458 hours	3.503 hours

Limits the authors state and we repeat: only three withanolide compounds were quantified, metabolism was not accounted for, and the participants were healthy Asian men. A plasma curve describes what is in the blood, not an effect on anyone.

Ashwagandha morning or night, and with or without food

There is no trial that gave one group ashwagandha in the morning and another the same dose at night, so the direct answer to “do I take ashwagandha in the morning or night” does not exist in the literature. What exists is a record of what each schedule was tested as.

The KSM-66 trials split the daily total into two servings, morning and evening, taken after food with milk or water. The Sensoril trial used one capsule at night, about half an hour before bed. The Shoden stress trial used one capsule after dinner with 250 ml of water. Taking ashwagandha before bed is therefore an ordinary schedule in the published work, not an experiment.

The practical logic follows from the pharmacokinetics rather than from any comparison. A high-glycoside extract with a mean residence time over 14 hours will still be measurable the next morning whichever end of the day you take it. A low-standardisation extract with a mean residence time near 3.5 hours will not, and splitting it across the day is the only way to keep any of it present for long. That is a reason to look at the standardisation figure before you decide the time, and it is the whole of what can honestly be said.

On food: the trials that specified it dosed after a meal. The published pharmacokinetics were measured fasting, so there is no human fed against fasted comparison to cite. Following the label and taking it with a meal is the schedule with the most trial-hours behind it.

Pick one time, keep it for the full 8 weeks, and do not change extract and schedule in the same month.

How long a course runs, and what happens when you stop

Most of the randomised evidence stops at 8 weeks. Langade 2019 ran to 10 weeks. Beyond three months the controlled data effectively runs out.

The longest published follow-up we could open is an observational study by Salve and colleagues (2026) in Phytotherapy Research, which followed 191 healthy adults aged 18 to 65 taking KSM-66 at 300 mg twice daily for 12 months, with laboratory assessments at baseline, 6 months and 12 months. Eighteen mild adverse events were reported and resolved without intervention, and alanine and aspartate transaminases did not change to a clinically significant degree. It is observational, uncontrolled, and therefore evidence about tolerability over time rather than about outcome.

On cycling, there is nothing to report. No published trial has compared continuous use against a cycled schedule, so any 5 days on, 2 days off rule you read was invented rather than measured. We are not going to invent one either.

On stopping: no trial we opened followed participants after withdrawal, so there is no published description of what happens when a course ends. If a page tells you otherwise, ask it for the study.

The practical position: treat 8 weeks as one complete trial of a product, then decide deliberately whether to continue, and speak to a pharmacist if you intend to keep going for many months.

What is not known, and who should not take it at any dose

The German Federal Institute for Risk Assessment published Mitteilung 039/2024 on 10 September 2024 and its position is stricter than the trial literature reads. Citing published risk assessments and internationally registered case reports, the BfR recommends against taking ashwagandha-containing preparations, and states that children, pregnant women, breastfeeding women and people with existing or previous liver disease in particular should not take them. It lists reported acute effects including nausea, vomiting, diarrhoea, dizziness, headache, drowsiness and skin rashes, and flags possible interactions with medicines for blood sugar, with antihypertensives and with immunosuppressants. It also notes that possible adverse effects have barely been studied in humans.

That position sits alongside trials reporting no adverse events, and both are true statements about different evidence. Small manufacturer-funded trials in healthy volunteers are not designed to detect rare harms. Case reports are.

Who should not take ashwagandha at any dose, on the BfR’s advice and ours: anyone pregnant or breastfeeding, children, and anyone with liver disease past or present. Anyone taking prescribed medication, in particular for blood sugar or blood glucose control, for hypertension, or any immunosuppressant, should speak to a doctor or pharmacist before starting. Anyone with a diagnosed condition should do the same. The side effects and regulatory picture gets its own article, because it deserves more than a paragraph.

Also unknown: no dose has been established for anyone outside the trial populations, most trials were funded by the extract manufacturers, and no trial has compared the extracts against each other for anything but plasma levels.

What to do with the label in front of you

The decision is not “how many milligrams”. It is three narrower questions. Which extract is this, and is it named. What is it standardised to, and on which analyte. What schedule was that extract actually tested on, one serving or two, and for how long. If the label answers all three, you can place your product against the trials above. If it answers none, the milligram figure on the front is a marketing number.

Our own ashwagandha and shilajit capsules hold 250 mg of ashwagandha extract and 250 mg of shilajit extract per capsule, and we are not going to print a withanolide percentage we have not had confirmed, because that is exactly the figure this page has just spent 2,000 words telling you to demand. If you want the reasoning behind pairing the two at all, that sits in ashwagandha and shilajit, and the wider category is covered in what adaptogens are.