Oxymetholone (Anadrol)
The strongest-bulking oral is one of the weakest androgen-receptor binders ever measured — and a direct, measurable estrogen-receptor agonist. Both of those sentences have receipts. Almost nothing else you've read about this drug does.
The short version
- Anabolic effect saturates at or below 100 mg/day in the only blinded RCT — 150 mg added zero gain over 100 mg and more liver toxicity. DATA
- It cannot aromatize (DHT-family structure) — yet it directly activates ERα at 562 nM, soy-isoflavone-tier potency. Aromatase inhibitors cannot block that. DATA
- Liver risk is dose- and duration-dependent; in a 110-patient cohort on ~150 mg, 15 of 17 people who quit, quit because of hepatitis. DATA
- No randomized trial exists in healthy trained men, in adolescents, or on growth plates. Anyone quoting you a number for those cases is extrapolating or guessing. DATA (an absence, verified by search)
Identity
17α-methyl-2-hydroxymethylene-DHT — fully 5α-reduced, A-ring saturated DATA
IUPAC skeleton in the binding literature: “17β-hydroxy-2-hydroxymethylene-17α-methyl-5α-androstan-3-one.” The 5α-androstan core is the DHT family — aromatase has no substrate to work on, so estradiol conversion from oxymetholone is structurally impossible, textbook steroid chemistry.
Does it work — the only real RCT
One blinded trial has ever compared oxymetholone doses head to head. It was run in HIV wasting — sick adults, on nothing else — and it was published twice from the same data:
| Arm | Dose | Weight gain (16 wk) | Liver toxicity |
|---|---|---|---|
| Placebo | — | — | reference |
| BID | 2 × 50 mg = 100 mg/d | +3.5 ± 0.7 kg | lower |
| TID | 3 × 50 mg = 150 mg/d | +3.0 ± 0.5 kg | higher |
1996 pilot: 30-wk open-label, n = 30 — peak weight +8.2 ± 6.2 kg (+14.5%) vs matched untreated controls losing 1.8 kg. Oxymetholone dose not stated in the abstract. DATA
Dose & return on investment
Above 100 mg/day, proven gain = zero DATA
The RCT's only dose comparison: 100 mg = 150 mg for weight, lean mass and body-cell mass. The 50→150 mg increment that separates the arms bought nothing measurable while liver toxicity rose.
Most of the effect is captured well below 100 mg MODEL
Assumption: anabolic dose-response is front-loaded (steep at low dose, flattening early) — the general shape for androgens, consistent with the flat 100-vs-150 datapoint. Prediction: 50 mg/d delivers most of the 100 mg effect. Never tested below 100 mg in a trial — this is inference, not measurement.
“25 mg works” CLAIM
Clinical anemia/HAE maintenance dosing lives at low doses, so biological activity is certain — but no trial quantifies the anabolic effect at 25 mg. Widely repeated, sourced nowhere.
Liver
17α-alkylated DILI spectrum: enzymes → cholestasis → peliosis → tumors DATA
Documented injury pattern for oral AAS: asymptomatic enzyme elevation through hepatocellular injury, cholestasis, fatty liver, vascular injury (peliosis) and neoplasia.
Real-world 150 mg: hepatitis is what ends treatment DATA
Cohort of 110 adults (aplastic anemia), initial oxymetholone 150 mg/day in 66.4%: 17.9% discontinued from side effects — 15 of those 17 from hepatitis. Response rate 56.4% overall.
Lipids
Oral androgens lower HDL-C, HDL particle number and size DATA
NMR lipoprotein profiling in androgen-treated patients (n = 9) vs untreated controls: significantly decreased HDL cholesterol, HDL particle number and HDL particle size — an atherogenic profile. Small n; direction consistent with the AAS literature.
Receptor pharmacology — the strange part
Androgen-receptor binding: too weak to measure DATA
Competition binding against radiolabeled methyltrienolone across a full AAS panel: nandrolone > methenolone > testosterone bound strongly; stanozolol/methandienone/fluoxymesterone were weak (RBA < 0.05); oxymetholone's RBA was too low to be determined. The strongest bulking oral is one of the weakest AR binders ever assayed.
…yet functionally it activates both AR and ERα DATA
NCATS/NCGC qHTS confirmatory reporter assays: AR agonism AC50 = 126 nM; ERα agonism AC50 = 562 nM — direct, with real concentration-response curves; antagonist modes inactive (~44.7 µM). Low radioligand-binding RBA and strong functional activation are different endpoints and are not contradictory.
The ERα number in context — same assay, one scale (agonist AC50, lower = stronger):
| Compound | ERα agonist AC50 | Reading |
|---|---|---|
| Diethylstilbestrol | 1.0 nM | the “real estrogen” benchmark |
| Genistein (soy isoflavone) | 398 nM | the “weak” benchmark |
| Oxymetholone | 562 nM | genistein-tier, ~560× weaker than DES |
| Testosterone (at ERα) | 12,589 nM | negligible |
What does not exist
No RCT in healthy trained men · none in adolescents · none on growth plates DATA
Verified by search (Europe PMC / ChEMBL, 2026-08-31): every controlled oxymetholone trial is in patient populations — HIV wasting, anemia syndromes, angioedema prophylaxis. No randomized trial in healthy trained subjects exists, no adolescent trial exists, and no controlled growth-plate data exists. Doses below 100 mg were never trialed. Any confident number for those cases is folklore wearing a lab coat.
Practical corollaries
An AI cannot block oxymetholone's estrogenic effects MODEL
Premise (data): the drug is non-aromatizable and agonizes ERα directly. Logic: aromatase inhibitors only reduce estradiol production; they do not occupy the receptor. Prediction: water retention or nipple symptoms on oxymetholone with a low estradiol blood reading is exactly what the pharmacology predicts — and more AI is the wrong tool for it. Levers that actually map: dose and duration.
Bloodwork stays interpretable on-cycle MODEL
Because oxymetholone contributes no estradiol, an E2 reading during a cycle reflects aromatizable compounds (e.g. testosterone) — the oral's estrogenic feel is receptor-level and invisible to that number. Monitor the drug by liver enzymes, lipids and blood pressure; monitor estrogen by the E2 assay.