SARMs: Eleven Names on One Shelf, and at Least Three Different Kinds of Thing

01What a SARM is supposed to be, and how many of these are one

Bond and colleagues, writing in Frontiers in Endocrinology in 2025, give the aim in one sentence: "The primary goal was to separate the unwanted androgenic or virilizing effects from the anabolic or myotrophic effects." Anabolic means tissue building, mostly muscle; virilizing means the effects that make a body more male, including on the prostate. Vignali and colleagues supply the chemistry in 2023, calling SARMs "non-steroidal compounds with favorable oral bioavailability", which is to say how much of a swallowed dose reaches the bloodstream. Two words there do the work: non-steroidal, and androgen receptor. Now go down the shelf. The World Anti-Doping Agency's 2026 Prohibited List names six of these as SARMs at section S1.2 on page 6, under a heading reading "Including, but not limited to" and introduced with "e.g.", so it gives examples rather than closing a set: andarine, enobosarm (ostarine), LGD-4033 (ligandrol), RAD140, S-23 and YK-11. Five pages later, at S4.4.1, under Metabolic Modulators, directly above insulins and meldonium, the same document reads "Rev-erbɑ agonists, e.g. SR9009, SR9011". Rev-erb is a protein that helps run the body clock, the daily rhythm of which genes are switched on and when. It is not the androgen receptor and it is not related to it.

GW-0742 is not named anywhere in that 26 page document, and a search of its extracted text for the string "0742" returns no occurrence. What the list does name at S4.4.1 is cardarine, printed as GW1516 and GW501516, as a PPAR delta agonist, a receptor involved in how cells handle fat and energy. GW-0742 is cardarine with one hydrogen swapped for a fluorine: PubChem gives cardarine as C21H18F3NO3S2 at 453.5 and GW-0742 as C21H17F4NO3S2 at 471.5. Same receptor as its sibling, and not the androgen receptor in either case.

YK-11 runs the other way. Anti-doping puts it in with the SARMs, while its founding paper, Kanno and colleagues at Toho University in 2011, opens with the words "A novel steroid compound", and PubChem's chemical name for it contains the string "cyclopenta[a]phenanthrene", the steroid nucleus itself. Both readings are on the record and this page is not going to quietly settle it.

02What is actually in the bottle

The most useful study here is not about any one of the eleven, and its sampling method decides what the numbers mean. A team funded by a United States Department of Defense health programme, through a grant to the United States Anti-Doping Agency, searched the internet over five weeks in early 2016 and found 210 products across 51 supplier sites. Forty-five could not be bought because the sites gave too little information, and 121 were out of stock, discontinued or restricted. That left 44 products from 21 suppliers, analysed between April and August 2016. The authors say plainly that the search was not exhaustive, that the results apply only to what they purchased, and that the sample is a single point in purchase time.

Twenty-three of the 44, or 52 percent, contained one or more selective androgen receptor modulator. The three the paper names in its own summary sentence are ostarine, LGD-4033 and andarine, and its product table adds a fourth: of three bottles sold as RAD140, one returned a formula match for it, one held ostarine and a trace of andarine instead, and one held nothing detectable. That is not the shape a shopper assumes: containing a SARM is not the same as containing the one on the label, and the labels across those products named at least seven different compounds. Another 17 held a different unapproved drug instead. Four held no active compound the laboratory could detect, and they cost 39.95, 54.99, 59.99 and 62.99 US dollars, for bottles labelled 5 mg of LGD-4033 per capsule, 25 mg of ostarine per capsule, 15 mg of RAD140 per capsule and 70 mg of a steroid-like plant extract per capsule.

The clearest single example is product 342084: sixty millilitres of solution at 129.99 dollars, labelled 50 mg per mL of ostarine. The analysis found 10 to 50 mg per mL of GW501516, plus ostarine below 0.1 mg per mL and tamoxifen below 0.1 mg per mL. The labelled compound was there in traces, a different one at something like the labelled strength, and tamoxifen, a prescription breast cancer medicine, turned up undeclared in four products in that sample. Eighteen of the 44 did match their label on both identity and strength, on a test generous enough that one bottle labelled 25 mg per mL was reported as 10 to 50 mg per mL and counted as a match.

Now the findings that cut the other way, which belong here rather than buried. A second team bought 13 products from two European websites in 2022, chosen on the Italian regulator's request that vendors be located in Europe: nine held the SARM on the label, three held a different one, one held none, and four held undeclared pharmaceutical substances including tamoxifen, clomifene, testosterone and tadalafil. Of the ten samples their quantitative method could measure, four came in at 80 to 95 percent of the label and three at roughly 60 to 80 percent. And a network of official medicines control laboratories looked at 324 samples across 13 countries over five years, most seized from illegal distribution rather than bought as test purchases, and wrote that from the limited quantitative data reported it can be assumed the majority contain active doses and some are even overdosed. So the honest summary is not that it is all fake. One buy-and-test study found half its sample was something other than advertised, a later one found most of its sample roughly right, and a five year seizure survey assumed most of its samples were dosed. Which of those describes the bottle in front of you is not knowable from any of them, and that is the finding.

Things a reader can check on a seller's own certificates page

• !

  One certificate per product, not one per batch

  One seller's page carries twelve entries for its SARMs and twenty for its peptides, one per product type, each showing only a date and a purity percentage. A certificate not tied to a batch cannot describe the bottle that ships.

• !

  No laboratory named on the certificate

  The same page names no issuing laboratory, no batch or lot number and no analytical method, while telling readers to 'call third-party labs to confirm that the results you see online are legitimate before purchasing products'. It names no laboratory to call.

• !

  The purity printed is below the purity promised

  That page states it tests every batch to a minimum of 98 percent, then publishes LGD-4033 at 97.29 percent and RAD-140 at 97.4 percent on the same screen. The seller's own document contradicts the seller's own floor.

• !

  A certificate whose issuing body would not issue one

  In the Italian sample, one retail site displayed an illegible certificate attributed to European Medicines Agency good manufacturing practices. Medicines agencies do not certify supplement purity.

03The scan number, the stair climb, and what the registry does not say

Ostarine is the only one of the eleven with a phase 3 trial in the registry searches behind this page, and it has two. POWER 1 and POWER 2 were run by GTx in people starting first line chemotherapy for non-small-cell lung cancer, giving enobosarm 3 mg by mouth once a day for 147 days against a dummy capsule; POWER 1 randomised 321 people and POWER 2 randomised 330, totals across both arms, with each trial set up as roughly 150 per arm. Both carried two co-primary endpoints, meaning two results that both had to come good, and both were scored as responder counts, an approach the design paper says was adopted "Based on extensive feedback from the US Food and Drug Administration (FDA)". A physical function responder had to improve stair climb power by at least 10 percent against their own baseline. A lean body mass responder had to show no loss of lean mass on a DXA scan, the X-ray scan that sorts a body into fat, bone and everything else.

At day 84, in POWER 1, 41.9 percent on ostarine held or gained lean mass against 30.4 percent on the dummy, and on the stair climb 29.4 percent against 24.2. In POWER 2, 46.5 percent against 37.9 on lean mass, and on the stair climb 19.5 percent against 24.8, which is the wrong direction. The registry gives confidence intervals around all eight figures and no statistical test of any kind: no analysis is posted against any of the four primary outcome measures, and the lean mass intervals overlap in both trials. So nothing here sourced to those records says an endpoint passed. What is on the record beyond them is that the programme missed its co-primary endpoints, and that thirteen years after the recorded primary completion in May 2013 the European register still shows no results filed against either protocol.

The hip fracture trial is the one with real statistics behind it, and the one that registered no functional outcome at all. Viking Therapeutics ran LGD-4033 under the code VK5211 in 108 people aged 65 and over who were three to seven weeks past a hip fracture, 28 on placebo and 29, 26 and 25 on 0.5, 1.0 and 2.0 mg, so 80 received the compound. Its single posted outcome measure is the primary one, placebo-corrected percentage change in lean mass on DXA at 12 weeks: 4.75 percent at the lowest dose, 7.15 at the middle and 9.08 at the top, with p equal to 0.0032 at 0.5 mg and below 0.0001 at the two higher doses. The record carries no secondary outcomes field at all. The six minute walk figure the company reports, about 22 metres over placebo, is the 2.0 mg arm alone and is called exploratory and not powered for significance on the company's own page.

What sits inside the scan number is why the gap matters. Twelve active men aged 18 to 29 cycled in the heat until they had shed 2.5 percent of their body mass, and their DXA lean tissue mass fell by 1.69 kg; two days of carbohydrate loading brought it back up by 2.36 kg, with no change in fat mass or bone mineral content at any timepoint. Two days of carbohydrate loading moved the lean compartment further than the 21 day phase 1 moved it at its top dose, where the average gain was 1.21 kg, and further than the 113 day cancer phase 2, where the median gain was 1.5 kg on 1 mg a day. Glycogen and the water bound to it live inside that number.

The half that cuts against all this belongs in the same section. In a 12 week double blind placebo controlled phase 2 in 120 healthy elderly men over 60 and postmenopausal women, randomised across five arms of 24 at 0.1, 0.3, 1 and 3 mg of ostarine and placebo, physical function was a secondary endpoint and it improved: P equal to 0.013 at 3 mg against placebo. That is the one trial of these eleven where the thing a person would notice moved, and the searches behind this page found no registration for it.

04Two files that are empty rather than clean

RAD-150 is where the naming goes strange. Look it up in PubChem with the hyphen and you get CID 32771, a haloalkylamine whose only appearance among those results is a 1975 paper on how such compounds behave in rat and rabbit blood. Look it up without the hyphen and you get two records, CIDs 68547459 and 68547462, the first of them "TLB 150 Benzoate", CAS 1208070-53-4, C27H20ClN5O3 at 497.9. That is RAD-140 with a benzoate group on its one esterifiable alcohol, and the arithmetic closes exactly: 393.8 plus 104.1 equals 497.9. So the molecule is drawn and registered, and everything downstream of the drawing is missing. A PubMed search for that CAS number returns nothing, a search on "RAD-150" or "RAD150" returns three papers from 1975, 1984 and 1994 on entirely unrelated subjects, and ClinicalTrials.gov returns nothing under any spelling.

LGD-3033 may not be a compound at all. A PubMed search for "LGD-3033" or "LGD3033" returns zero records, and the only database entry the name produces is PubChem's record for LGD-3303, one digit away, where "LGD-3033" sits among the depositor-supplied synonyms. LGD-3303 itself is real: a PubMed search returns five records, and the three the research opened are all in rats, two of them out of Ligand Pharmaceuticals. And people who ought to know treat the two spellings as separate things: a European medicines control study lists both in its scope, the American Department of Defense watch list spells it LGD-3303, and a 2019 Senate bill enumerated LGD-3303 by its full chemical name.

That is why an empty file and a clean file have to be kept apart. Nothing published says these two are harmful. Nothing published says anything about them.

05The pharmacology that ends one of them

SR-9009's problem is not mainly toxicological. It is that the compound barely survives being swallowed.

The paper establishing this is the one that set out to improve on its chemistry, a GlaxoSmithKline group publishing optimised chemical probes for Rev-erb alpha. They ran the compounds through mice, three animals per time point, nine per dosing route, at 1 mg per kilogram. SR-9009's oral bioavailability came out at 2.2 percent, and its terminal half-life, measured on the intravenous side of the same table, was 0.55 hours, about 33 minutes. Their own sentence about what that implies is the one to read: these compounds "would be suitable for acute time-of-day dosing by injection where short exposure of the compound at meaningful levels in desired".

The two famous mouse results obeyed that. In the 2012 Nature paper, obese mice on a high fat diet got SR9009 at 100 mg per kilogram twice a day for 30 days, injected into the abdominal cavity, six to ten animals per group, and lost weight against vehicle-treated mice. In the 2013 endurance paper, the mice that ran longer got 100 mg per kilogram for 30 days by the same route, six per group. Not a capsule anywhere.

There is a friendlier bioavailability figure in circulation, around 24 percent, attributed to that same study. It belongs to a different molecule. The bioavailability row of that table reads 2.2, 23.4, 2.4, 3.5 and 3.0 percent across five compounds, and the 23.4 is compound 4, GSK2945, singled out by the authors for a longer half-life of 2.0 hours. Every figure in the table was measured in mice, so none is a human number. What circulates on the retail side is 10 to 40 mg a day depending on experience, reported as convention and not as a recommendation, with an instruction to redose every two to four hours because the compound does not last. That last instruction is the only place in this whole literature where a vendor page and a pharmacology paper agree with each other.

06Where the doses come from

1

2

3

4

1

10

100

mg per day · log scale

derived from researchcommunity convention

1. 1Top arm, 21 day phase 1 in 76 healthy men · 1 mg per day1.0 mg daily for 21 days, the highest dose in the placebo controlled phase 1 in 76 healthy men aged 21 to 50. Lean body mass rose by an average 1.21 kg at that dose, p equal to 0.047 against placebo.
2. 2Top arm, 12 week phase 2 in hip fracture patients · 2 mg per day2.0 mg daily for 12 weeks, the highest arm of the 108 person hip fracture phase 2. Twenty-five people took that dose.
3. 3The figures on one guide page. Convention, not a recommendation · 10 to 15 mg per dayOne guide page states 10 mg a day for men over an 8 to 10 week cycle, 5 mg for women and up to 15 mg for experienced users. Reported as convention, not as a recommendation, and describing one website rather than the market.
4. 4What that page attributes to clinical trials. Convention, not a recommendation · 222 mg per dayThe same page tells readers that 'According to results obtained from clinical trials, humans can take up to 222 mg per day of Ligandrol for 14 consecutive days.' Its citation list has exactly two entries, and the only trial among them is the 21 day phase 1 above, whose highest dose was 1.0 mg. Reported as convention, not as a recommendation.

RAD-140 runs the other way, and this is the one place where reading the LGD-4033 pattern across would produce a false sentence. The completed first-in-human trial gave 50 mg a day to six patients, 100 mg to thirteen and 150 mg to three, and set the maximum tolerated dose at 100 mg a day. What circulates for RAD-140 is 20 to 30 mg a day for men on an eight week cycle, reported as convention, and it sits below what was studied rather than above it. The page publishing those figures cites no human study for them.

07What could go wrong, and a correction this page carries

First the correction. Until today the index on this site told readers that these eleven compounds carry documented liver toxicity across the class. That is not what the record shows, and the index has been changed. It is worth saying plainly, because it is the same error the sellers make in the opposite direction: treating eleven different things as one thing, and moving a finding from one of them onto all of them.

Here is what is supportable. Published case reports naming a single compound as the cause attach to three of the eleven: LGD-4033, ostarine and RAD-140. Three more, YK-11, S-23 and andarine, appear only inside mixed exposures where somebody took two or three compounds at once, or as counts in the FDA's adverse event database for foods and supplements, which are reports somebody submitted rather than adjudicated findings. For four names, RAD-150, LGD-3033, AC-262 and GW-0742, the searches behind this page found no published human harm record at all. SR-9009 is the one the searches disagreed on: a PubMed search for it alongside liver terms returns four records, one a 2025 case report of a 40 year old man with liver injury after starting stenabolic, which the authors judged the likely source after ruling out infection and autoimmune disease. He was taking several other over-the-counter supplements, no product was analysed, and one case is a weak signal.

None of the eleven holds a medicines approval anywhere named on this page, and the Department of Defense's Operation Supplement Safety puts that plainly: "To date, no SARM has been approved by the Food and Drug Administration (FDA) for any medical use." What the American regulator has instead is warning letters, and its sentence in them is stronger than the one this site had, not weaker. In every one of the twelve this research read, from October 2017 to December 2025, the same sentence appears, unhyphenated in the 2017 letters and hyphenated in the 2025 ones. Infantry Labs, October 2017: "Life threatening reactions, including liver toxicity, have occurred in people taking products containing SARMs." Products. Never a named compound, not once in twelve letters. That turns out to be the right unit, because only 52 percent of products sold as SARMs in the 2017 analysis contained one. A warning about products is a warning about a supply chain, which is what the evidence supports.

Six of the eleven are named somewhere across those twelve letters: ostarine, LGD-4033, RAD-140, andarine, S-23 and YK-11. RAD-150, LGD-3033, AC-262, SR-9009 and GW-0742 appear in none of them.

The individual cases are severe. A 24 year old man took RAD-140 by mouth at up to 15 mg daily for five weeks and reached a total bilirubin of 38.5 mg per dL, roughly thirty times the top of normal; thirteen weeks after stopping his bilirubin was still 4.9 and his ALT still 118, and he did not come back for his next appointment. But an Australian series of 23 people hospitalised with drug induced liver injury supplies the number that should govern all of it. Those 23 had taken 40 drugs between them, 18 anabolic steroids, 14 SARMs and eight other bodybuilding supplements, so the 14 is a count of exposures and not of patients. Under a formal causality method the injury was rated possible in 17 of the 23, probable in 2 and unlikely in 4.

The counterweights are real. Seventy-six healthy men aged 21 to 50 were randomised to a dummy or to 0.1, 0.3 or 1.0 mg of LGD-4033 daily for 21 days, and neither liver enzyme changed significantly at any dose; what fell instead was HDL cholesterol, triglycerides, total testosterone and the protein that carries it, all in proportion to dose and all back to baseline after the men stopped. In the hip fracture phase 2 the liver terms cluster at the top dose and none was classed as serious. And among 899 cases of drug induced liver injury enrolled in a United States prospective study between 2004 and 2013, a National Institutes of Health resource notes, none were attributed to SARMs.

08What I could not settle

Whether the two phase 3 trials passed or failed their lean mass endpoint on the registry's own evidence. It posts responder percentages, overlapping confidence intervals and no statistical test.

Whether registry silence means anything. Viking's own page describes VK5211 as having run in three phase 1 trials, and only one VK5211 study exists on ClinicalTrials.gov, so eight of eleven returning nothing is a statement about what is registered, not about what was run.

Whether andarine does anything to vision, which is the most repeated claim about any of these compounds. The searches behind this page found one report of blindness in an adverse event database, filed under the generic keyword rather than under andarine, in a table showing only ten of the fifteen reports that keyword returned.

Whether YK-11 inhibits myostatin in a person. The claim rests on a 2013 study in which YK11 induced follistatin, the protein that binds myostatin, in mouse muscle cells in a dish. One animal study exists, in bacterial sepsis. Nothing in a person turned up.

09What the shelf adds up to

The most useful fact on this page is a fact about a document, and it costs nothing to check. Open the World Anti-Doping Agency's 2026 Prohibited List, a free 26 page PDF. Go to page 6 and read section S1.2, where six of these compounds are named as selective androgen receptor modulators. Then go to page 11 and read S4.4.1, where SR9009 sits as a Rev-erb agonist under metabolic modulators, directly above the entries for insulin and meldonium. Then search the whole document for the string "0742" and watch it return nothing.

Five pages, in a document written by people whose job is to know exactly what a substance does, separate SR-9009 from the six the same document names as SARMs. GW-0742 is not in the document at all. A shelf that puts them together is not making a scientific claim. It is making a merchandising decision, and that decision is what most of the writing about these compounds inherits, including the sentence this site had published until today.

If you have looked at a certificate of analysis for one of these, did it name the laboratory that issued it, and did it carry the lot number printed on your bottle?

Frequently asked

Are SARMs approved for anything?+

No medicines approval for any of these eleven turned up in the searches behind this page. The United States Department of Defense's supplement safety programme states it directly: to date, no SARM has been approved by the Food and Drug Administration for any medical use. FDA's own consumer page says SARMs are considered unapproved drugs and cannot be legally marketed in the United States as a dietary supplement or a drug at this time. UK Anti-Doping reports that the Food Standards Agency treats them as unauthorised novel foods, which makes them illegal to sell or advertise in food or supplements in the United Kingdom.

Do SARMs cause liver damage?+

It depends entirely on which compound, and the honest answer is not a class answer. Published case reports naming a single compound as the cause attach to three of the eleven: LGD-4033, ostarine and RAD-140, several of them severe, with jaundice and bilirubin many times the top of normal. Three more appear only where somebody took two or three compounds at once, or as counts in an adverse event database. For four of the eleven the searches behind this page found no published human harm record at all. In all twelve FDA warning letters this research read, the agency's liver sentence is about products containing SARMs and never about a named compound, which fits a 2017 analysis in which only 52 percent of products sold as SARMs actually contained one. Against that, 76 healthy men were randomised to LGD-4033 or a dummy for 21 days, and neither of the two liver enzymes on a routine blood test changed significantly at any dose.

Is what is in the bottle the compound on the label?+

Two buy-and-test studies disagree. In 44 products bought online in 2016 from 21 suppliers, 23 contained a selective androgen receptor modulator of any kind, 18 matched their label on both identity and strength, four contained no active compound at all, and one bottle sold as ostarine at 129.99 dollars contained a different compound at label-like strength with the labelled one present only in traces. In 13 products bought in Italy in 2022, nine held the SARM on the label, three held a different one and one held none. A five year survey of 324 seized samples across 13 countries assumed from limited data that most contained active doses and some were overdosed. None of the three tells you about the bottle in front of you.

Are SR-9009 and cardarine SARMs?+

No, and the anti-doping list says so by where it puts them. The 2026 Prohibited List names six SARMs at section S1.2 on page 6 and lists SR9009 five pages later at S4.4.1 as a Rev-erb agonist under metabolic modulators, alongside insulins and meldonium. Rev-erb is a body clock protein, not the androgen receptor. Cardarine, printed there as GW1516 and GW501516, appears in the same section as a PPAR delta agonist, a receptor involved in fat and energy handling. GW-0742 is not named anywhere in that 26 page document; it is cardarine with one hydrogen swapped for a fluorine. A resource from the National Institutes of Health groups stenabolic and cardarine among products 'sometimes mistakenly claimed to be SARMs'.

Did any of these actually work in a trial?+

Three of the eleven have a registered human trial in the searches behind this page. In the two phase 3 trials of ostarine, both carried two co-primary endpoints at day 84, a body scan measure and a stair climb measure. The scan measure showed more responders on the drug in both trials. The stair climb measure showed a smaller gap in the same direction in the first trial, 29.4 percent against 24.2, and went the wrong way in the second, 19.5 against 24.8. Those registry records post percentages and confidence intervals and no statistical test, so nothing here says an endpoint passed. A phase 2 in 108 hip fracture patients did post statistics and did increase lean mass by 4.75 to 9.08 percent over placebo, and its registry record carries no functional outcome at all. The one trial where a functional measure improved against placebo was a 12 week phase 2 in 120 healthy elderly volunteers, at 3 mg of ostarine, at a P value of 0.013, meaning a gap that size would turn up by chance about thirteen times in a thousand runs if the drug did nothing.

What about RAD-150 and LGD-3033?+

Under any spelling the research tried, both return nothing published in a person. RAD-150 does have a drawn structure and a CAS number, 1208070-53-4, as the benzoate ester of RAD-140, but a PubMed search for that number returns nothing, there is no registry record, and no published analysis of what is sold under the name. LGD-3033 has no record of its own at all; the name appears only among depositor-supplied synonyms on the PubChem entry for LGD-3303, a different compound one digit away, studied only in rats. That is an empty file rather than a clean one. Nothing published says they are harmful, because nothing published says anything about them.

---

This content is for educational and informational purposes only and is not medical advice. Compounds discussed are research compounds and are not approved for human use. Nothing here should be used to diagnose, treat, cure, or prevent any disease. Always consult a qualified healthcare professional before starting any compound, supplement, or protocol. Individual responses vary. Do not self-administer compounds without proper medical supervision.