What Does the Science Say About Shilajit? The Clinical Trials, Explained

Every human trial, at a glance

These are the published studies that gave shilajit to people. Animal and cell studies — a much larger literature — are covered further down, because they are routinely quoted as though they were human results.

Testosterone: the strongest claim, and its limits

Pandit 2016 is the trial that testosterone marketing rests on, and it is a legitimate one: randomised, double-blind, placebo-controlled, 90 days, 250 mg of purified shilajit twice a day in healthy men aged 45 to 55. Total testosterone, free testosterone and DHEAS all rose significantly against placebo, and the gonadotropins that would signal a disrupted feedback loop stayed where they were.

Two qualifications matter. The population was middle-aged men, chosen because testosterone declines with age — the result does not transfer automatically to a 25-year-old with normal levels. And it is a single trial that has not been replicated by an independent group.

The most useful outside check comes from a 2024 systematic review in the International Journal of Impotence Research, which assessed two decades of evidence on 27 marketed testosterone boosters across 52 studies. Its finding was that most of them fail to raise total testosterone at all. Purified shilajit extract was one of a short list rated "possibly effective" — and specifically for men with late-onset hypogonadism, not for men in general. Being one of the few ingredients that survives that review is a meaningful distinction in a category full of ingredients that do not.

A related question this evidence does not answer is erectile function. No trial has tested shilajit against ED, and the paper most often cited as if it had does not contain shilajit — we go through that in detail in Shilajit and Erectile Dysfunction.

Fertility: a real signal from a weak design

Biswas 2010 reported large improvements in oligospermic men — sperm count up 61.4%, testosterone up 23.5%, oxidative stress in semen down — over 90 days at 100 mg twice daily. The effect sizes are striking and the safety data (unchanged liver and kidney profiles) is useful.

But the design was open-label with a single arm: every participant knew they were taking shilajit, and the comparison was to their own baseline rather than to a placebo group. Semen parameters vary substantially within the same man between samples, and participation in a fertility study changes behaviour. This study is frequently described online — including, until we corrected it, elsewhere on this site — as a randomised placebo-controlled trial in 96 men. It was neither. Twenty-eight men completed it, and there was no control group.

Muscle, collagen and bone

Four studies form a coherent thread here, and it is not the thread marketing usually pulls. Das 2016 found extracellular-matrix genes — collagen, elastin, fibronectin — were upregulated in muscle biopsies after 8 weeks. Neltner 2024 found a circulating marker of type 1 collagen synthesis roughly doubled at 500 and 1000 mg/day, where placebo did not move. Das 2019 found new blood-vessel and matrix genes induced in skin, with measurably better microperfusion. And Sadeghi 2020 — the largest trial, 160 patients recovering from tibia fractures — found bone union about 24 days faster than placebo.

Taken together these point at connective tissue and repair rather than at energy or strength. Keller 2019 fits the same frame: participants taking 500 mg/day did not get stronger, they lost less strength across a fatiguing protocol. Read the fine print there though — that effect appeared in the upper half of the sample by baseline strength, not in the group as a whole, and not at the 250 mg dose. It is a subgroup finding in a trial of 63 people, which makes it a lead worth following rather than a settled result. Our muscle recovery guide goes through what "fatigue resistance" meant in practice.

One caution on Sadeghi 2020: it used momiai, the Iranian form of mineral pitch, which is not necessarily the same standardised extract used in the US trials. Shilajit is a natural material whose composition varies by region, and trials on one preparation do not automatically describe another.

The two newest studies, and why neither settles anything

Two studies published since 2023 are worth separating out, because both address questions buyers actually ask and both have designs that cannot answer them.

Yadav 2026 is the only published human study of shilajit resin rather than a standardised extract, which makes it genuinely novel — most of the market sells resin and almost none of the research has used it. Over 28 days, 25 healthy men improved on leg press, endurance, grip strength and VO₂ max, and their fatigue scores fell by about a third. It is also open-label, single-arm and run by the manufacturer: everyone knew what they were taking, there was nobody to compare them to, and the headline outcome was a questionnaire. People who sign up for a supplement study, take it for a month and are then asked whether they feel less tired tend to say yes. Read it as a safety and feasibility signal — which is what a pilot study is for, and what its authors say it is.

Patil 2023 is the closest thing to evidence on shilajit and blood pressure. Sixty elderly hypertensive patients took 500 mg twice daily on top of their antihypertensive medication for 30 days, against controls who took the medication alone. Oxidative stress markers moved in shilajit's favour. Arterial stiffness and endothelial function — the measures that would plausibly translate into blood pressure — did not change. Anyone citing this study as evidence that shilajit lowers blood pressure is citing a result it did not report.

Five reasons to hold this evidence loosely

• The trials are small. Eight of the ten enrolled fewer than 70 people. Small trials overestimate effects, and a single unreplicated result in 35 or 63 participants is a hypothesis with good manners.
• They mostly test one product. A large share of this literature — including Keller 2019, Das 2016 and Das 2019 — used PrimaVie, a single branded extract, and the manufacturer appears in the author list of at least one of them. That is normal for supplement research and it is still a reason to want independent replication. It also means the results describe that preparation, not shilajit in general.
• Several measure markers, not outcomes. Gene expression in a biopsy and a collagen synthesis marker in serum are mechanistic evidence. They are not the same as stronger tendons, better skin or faster recovery, and the step from one to the other is exactly where supplement marketing tends to jump.
• Some have no control group. Biswas 2010 and Das 2016 compared participants to their own baselines. That design cannot separate the supplement from time, expectation or the attention of being in a study.
• Nothing here runs long. The longest trial lasted 14 weeks; the two most recent ran 28 and 30 days. Shilajit is marketed for daily, indefinite use. Nobody has studied what a year of it does.

Where the animal studies get quoted as human ones

The animal and cell-culture literature on shilajit is several times larger than the human one, and it is where most of the mechanistic claims come from. That is legitimate research. The problem is the citation drift: a rat result is described in a blog post without the species, gets repeated, and becomes "a study showed".

The clearest example is the chronic fatigue work. Surapaneni 2012 found that shilajit reversed fatigue behaviours and protected brain mitochondria in a forced-swim model — in rats, at 25 to 100 mg per kilogram of body weight, for 21 days. It is regularly cited online as a double-blind placebo-controlled study in chronic fatigue syndrome patients at 200 mg twice daily. There is no such study. A dose of 100 mg/kg in a rat is also not a human dose: naive scaling would put it in the grams for an adult, far above anything any human trial has used.

The same applies to the blood-sugar, cognition and immune claims: the diabetes work is in alloxan-induced diabetic rats, the Alzheimer's material is mechanistic and largely preclinical, and the immunomodulation findings come from fulvic acid in cell culture. None of that makes the claims false. It makes them untested in people.

What nobody has studied

The gaps are as informative as the findings, because several of them sit under the best-selling claims on the market:

• Energy in healthy adults — no controlled trial. The single study measuring fatigue in healthy people (Yadav 2026) had no placebo group and no blinding, which is the worst possible design for a self-reported outcome. This is the most common reason people buy shilajit.
• Cognition, focus and memory in humans — no trial with clinical endpoints.
• Sleep — no trial; the mechanism is extrapolated from animal HPA-axis data.
• Women, for anything hormonal — the only women's trial measured skin.
• Form comparisons — no trial has compared resin against capsules, gummies or tablets for anything. Only one study has used resin at all, and it had nothing to compare against.
• Blood pressure — the one trial in hypertensive patients measured oxidative stress, found no change in arterial stiffness or endothelial function, and was an add-on to medication rather than a substitute for it.
• Long-term use, children, pregnancy — untested, and in the last two cases for good reason. See our safety guide.

What this means when you are buying

Three practical things fall out of the trials. First, the dose: every study that found anything used 250 to 1000 mg per day of a purified, standardised preparation, usually split into two. A product delivering 50–200 mg of extract per serving is not delivering what was studied, which is part of why our grading caps the heavily diluted formats.

Second, the duration: results appeared at 8 to 14 weeks, not in the first fortnight. Third, and most important, the material: these trials used purified, standardised, and in most cases laboratory-characterised shilajit. Nothing in this literature describes the safety or the effect of an untested resin of unknown origin. The evidence such as it is applies to products that can document what they are — which is what the third-party tested list exists to identify.