Cordyceps

01 What this document is, and four things it is not

This article is a research review of the commercial name Cordyceps: the wild caterpillar fungus now classified as Ophiocordyceps sinensis, cultured mycelium sold as if it were that fungus, Cordyceps militaris fruiting bodies and mycelia, fermentation isolates, and the nucleosides used as stand-ins for identity. It is written against a market that treats a Tibetan prestige object, a Chinese fermentation brand, an orange club fungus, and a 3'-deoxyadenosine peak as interchangeable evidence.

Four things follow immediately.

First, this is not a foraging or cultivation manual. Wild O. sinensis is a high-altitude insect-fungus complex with a conservation and adulteration problem (Lo et al., 2013; Li, Yang, and Tsim, 2006). Nothing here is a harvest instruction.

Second, it is not a licence to read traditional kidney and lung claims as modern trial results. Dong chong xia cao is a traditional indication set. Cochrane's CKD review of Chinese preparations is a different object, and its authors refused a definitive conclusion (Zhang et al., 2014).

Third, it is not an athletic protocol. The load-bearing human exercise file is small, product-specific, and split: a trained-cyclist null, an elderly threshold signal without a VO2max change, a mushroom-blend exploratory cell, and two herbal blends that cannot isolate the fungus (Parcell et al., 2004; Chen et al., 2010; Hirsch et al., 2017; Earnest et al., 2004).

Fourth, it is not medical advice. This document recommends no use, dose, route, or schedule for any person.

02 Taxonomy: the genus that would not stay one genus

Until 2007, the commercial sentence "Cordyceps sinensis" could pretend to be a genus fact. Sung, Hywel-Jones, Sung, Luangsa-ard, Shrestha, and Spatafora analysed 162 taxa across five to seven loci and rejected the monophyly of Cordyceps as then defined. They validated Cordycipitaceae on the type Cordyceps militaris — brightly coloured, fleshy stromata — and proposed Ophiocordycipitaceae for dark, tough stromata, moving the Chinese caterpillar fungus into Ophiocordyceps (Sung et al., 2007). That revision is established as phylogenetic taxonomy. It is not a marketing inconvenience to be ignored on a label.

The practical consequence is prior to every later section. C. militaris and O. sinensis are not sister products in one family. A paper written about one is not evidence about the other unless the authors authenticated the material. Reviews that still say "Cordyceps, family Clavicipitaceae, more than 400 species" after 2007 are using a pre-split map (Yue et al., 2013). Yue and colleagues remain useful as a chemical catalogue. They are not a current taxonomic warrant.

03 Wild Ophiocordyceps sinensis: the prestige object

The wild material is an insect-fungus complex: a Hirsutella anamorph and an Ophiocordyceps teleomorph on ghost-moth larvae of the Tibetan plateau, sold in Chinese as dong chong xia cao, "winter worm, summer grass" (Chen et al., 2001; Lo et al., 2013; Li, Yang, and Tsim, 2006). Lo, Hsieh, Lin, and Hsu's systematic review of that complex treated the organism, the substitutes, and the bioactive-ingredient literature as a single messy file, not as a settled nutraceutical (Lo et al., 2013). That review is strongly supported as a map of confusion. It is not a trial.

Traditional use is kidney and lung replenishment, fatigue, and a long list of other indications that Li, Yang, and Tsim recited as the reason quality control exists (Li, Yang, and Tsim, 2006). Reciting a traditional list is established as ethnopharmacology. Extrapolating it to a Western supplement aisle is speculative. Panda and Swain's Sikkim paper is a regional-use note, not a clinical programme (Panda and Swain, 2011).

Price, scarcity, and prestige are the industrial facts that created everything else in this article. When the wild complex is rare, fermentation products and C. militaris occupy the shelf, and adenosine or cordycepin occupy the certificate of analysis (Li, Yang, and Tsim, 2006; Yue et al., 2013). Paterson, reviewing the genus as a "fungal therapeutic biofactory," already stressed that laboratory catalogues had outrun human trials (Paterson, 2008). LiverTox, updating the hepatotoxicity chapter in June 2025, repeated the same gap in plainer language: laboratory and animal findings have not become convincing clinical effects (LiverTox, 2025).

04 Cultured mycelium, Cs-4, and the anamorph that is not Hirsutella

Most commercial "Cordyceps sinensis" is not the wild complex. It is a cultured mycelium or a fermented mycelial powder. Li, Yang, and Tsim said so in 2006 as a quality-control premise, not as an accusation (Li, Yang, and Tsim, 2006). LiverTox said so again in 2025: most current supplements are a laboratory-grown version referred to as Cs-4 (LiverTox, 2025). That industrial fact is established.

The taxonomic fact sitting under it is sharper. Chen, Wang, Qu, Li, and Zhang sequenced ITS and 5.8S from plateau C. sinensis teleomorphs and several asexual isolates. Distance to Hirsutella sinensis was below 0.02. Distance to Paecilomyces sinensis was 0.34. They concluded that H. sinensis is the anamorph, and that Paecilomyces isolates obtained in culture are different organisms (Chen et al., 2001). Ko, Liau, Lee, Chiu, Martel, Lin, and colleagues later isolated and characterised H. sinensis mycelium from the fruiting-body complex, which is the culture that would actually be the wild fungus if a manufacturer grew it (Ko et al., 2017). A 2008 phylogenetic analysis of Paecilomyces hepiali and C. sinensis exists because the two names had already been stacked in commerce (Yang et al., 2008).

Cs-4, Bailing, Jinshuibao, and CBG-CS-2 are therefore brand and strain names, not synonyms of wild O. sinensis. Jung, Jung, Choi, Sin, Ha, and Chae's immune trial labelled the product "Cordyceps" in the title and Paecilomyces hepiali CBG-CS-2 in the methods (Jung et al., 2019). That is not a quirk. It is the identity problem written into a randomised trial. Grade of "Cs-4 equals O. sinensis": speculative, and rejected here.

05 Cordyceps militaris: the species that can be grown

C. militaris is the type of Cordyceps in the split taxonomy (Sung et al., 2007). It produces orange stromata, can be cultivated, and is the organism from which cordycepin was isolated as a metabolic product (Tuli, Sandhu, and Sharma, 2014; Ashraf et al., 2020). Those are established mycological and chemical facts. They do not make C. militaris a cultivated O. sinensis.

The market shift toward C. militaris is rational on supply grounds and dishonest if the label still borrows Tibetan prestige. Li, Yang, and Tsim already listed C. militaris among substitutes and adulterants as a reason to do analytical chemistry (Li, Yang, and Tsim, 2006). Olatunji, Tang, Tola, Auberon, Oluwaniyi, and Ouyang's later genus review and Das and colleagues' immune-potential review both keep the two species in one narrative more tightly than the phylogeny warrants (Olatunji et al., 2018; Das et al., 2020). Those reviews are useful inventories. They are not licences to pool trials.

Hirsch, Smith-Ryan, Roelofs, Trexler, and Mock did not even test isolated C. militaris. They tested PeakO2, a mycelial biomass powder cultured on organic oats plus other mushroom species, dosed as 4 g per day of "mushroom blend" (Hirsch et al., 2017). That trial is taken up in section 10. It belongs here first as an identity exhibit: even the paper that the aisle cites for "Cordyceps militaris and VO2max" is a blend on grain.

06 Cordycepin is a 3'-deoxyadenosine, not a species proof

PubChem records cordycepin as CID 6303, molecular formula C10H13N5O3, molecular weight 251.24, IUPAC name (2R,3R,5S)-2-(6-aminopurin-9-yl)-5-(hydroxymethyl)oxolan-3-ol, InChIKey OFEZSBMBBKLLBJ-BAJZRUMYSA-N (PubChem, 2026a). Adenosine is CID 60961, C10H13N5O4, 267.24, the same purine with a 3'-hydroxyl still in place (PubChem, 2026b). The structural difference is one oxygen. That database contrast is established. It is the whole of the "ATP-related" marketing story that is chemically true.

Ashraf and colleagues stated the same difference and then listed a nutraceutical catalogue — anti-diabetic, anti-aging, antiviral, cosmeceutical — as if the missing hydroxyl licensed the list (Ashraf et al., 2020). The chemistry is established. The catalogue is speculative as human outcome evidence. Tuli, Sandhu, and Sharma's earlier cordycepin review is a pharmacological inventory of the same kind (Tuli, Sandhu, and Sharma, 2014). Inventories are not trials.

Cordycepin is abundant in many C. militaris fermentations and often low or unstable in wild O. sinensis material; nucleoside profiles also move with sample preparation (Li et al., 2001; Fan et al., 2008; Yue et al., 2013). Standardising a sinensis-labelled powder to cordycepin can therefore select for the wrong species or for a spiked isolate. That is a quality argument, not a health argument.

07 Adenosine, mannitol, polysaccharides, ergosterol

Adenosine is a genuine constituent of many Cordyceps preparations and a genuine human signalling nucleoside. Those two facts do not stack. Chen and colleagues' Cs-4 powder was listed at 0.14% adenosine among adenine, uracil, uridine, 5% mannitol, 0.5% polysaccharides, amino acids, and micronutrients (Chen et al., 2010). That is a manufacturer's composition list in a trial methods section. It is established as what that paper said was in the capsules. It is not a plasma PK study, and it is not a proof that oral adenosine from the powder reaches muscle mitochondria.

D-mannitol is CID 6251, C6H14O6, 182.17 (PubChem, 2026c). Older literature called it "cordycepic acid." Mannitol is a sugar alcohol. It is not a unique Cordyceps toxin or a unique Cordyceps tonic. Using a mannitol peak as identity is speculative.

Polysaccharides are the other favourite marker class. They are structurally heterogeneous, assay-dependent, and routinely asked to carry immunomodulatory claims that the human file does not support at the infection-outcome level (Yue et al., 2013; Das et al., 2020; Liu, Wang, Wang, Zhang, Zhang, and Han, 2015). Ergosterol and fatty-acid profiles distinguish fungal biomass from plant starch to a degree; Yang, Feng, Zhao, and Li used a one-step derivatisation GC-MS method on natural versus cultured material for that reason (Yang et al., 2009). A sterol chromatogram is emerging as a biomass marker. It is not a clinical endpoint.

Li, Li, Dong, and Tsim compared anti-oxidation activity across natural C. sinensis and cultured mycelia and found large source variation (Li et al., 2001b). Variation is the finding. "Antioxidant, therefore health" is the inference this article refuses.

08 Why a nucleoside peak is not identity

Fan, Yang, Duan, Li, Chen, and Li showed that sample preparation changes quantified nucleosides in natural and cultured Cordyceps (Fan et al., 2008). Li, Li, Dong, and Tsim had already compared nucleoside contents by capillary electrophoresis between natural C. sinensis and cultured mycelia (Li et al., 2001). Together those papers are strongly supported as analytical warnings. They are the reason a single adenosine or cordycepin specification cannot authenticate a species, a tissue (fruiting body versus mycelium), or a substrate (grain versus insect).

The commercial move is the reverse: pick a marker, print it, and imply the organism. Cs-4's 0.14% adenosine in Chen and colleagues is the clean example (Chen et al., 2010). PeakO2's oat-cultured mycelial biomass in Hirsch and colleagues is the other (Hirsch et al., 2017). Grain mycelium will always pass a "fungal powder" inspection that does not ask how much of the mass is leftover starch. That problem is taken up in section 16.

09 Traditional use is not a trial

Zhu, Halpern, and Jones's 1998 "scientific rediscovery" review is the prestige document of the English-language supplement era. It surveyed English and Chinese preclinical work and "blinded or open-label trials in to date over 2000 patients" and listed radical scavenging, antisenescence, endocrine, lipid, and sexual-function activities (Zhu, Halpern, and Jones, 1998). That paper is established as a historical review. It is not a randomised evidence synthesis. Paterson later said the clinical studies had not demonstrated the biologic effects in humans (Paterson, 2008). LiverTox in 2025 is still saying they have not (LiverTox, 2025).

The traditional kidney/lung frame explains why Chinese hospital products exist. It does not licence a VO2max claim in a trained cyclist, an NK-cell claim in a healthy Korean man, or an aging claim in a longevity catalogue. Grade of traditional-use extrapolation to those endpoints: speculative, and rejected in section 20.

10 Exercise, VO2max, and fatigue

Four human exercise objects matter. They are not one literature.

Parcell, Smith, Schulthies, Myrer, and Fellingham (2004) gave CordyMax Cs-4, 3 g per day for five weeks, to 22 endurance-trained male cyclists. VO2peak was 59.9 versus 59.1 ml/kg/min before supplementation and 60.1 versus 57.1 after; ventilatory threshold stayed at 72% of VO2peak; time trials were 61.4 versus 62.1 minutes. Nothing moved (Parcell et al., 2004). That paper is strongly supported as a null in trained men at 3 g per day of that product. It is the trial the aisle least likes to quote.

Chen, Li, Krochmal, Abrazado, Kim, and Cooper (2010) enrolled 20 healthy adults aged 50–75, randomised 1:1 to Cs-4 (CordyMax) or starch placebo for 12 weeks, and analysed 15 completers (8 Cs-4, 7 placebo). The methods and the 756-capsule issue imply nine 333 mg capsules per day, about 3 g; the abstract's "333 mg … 3 times a day" does not match that count. After 12 weeks, metabolic threshold in the Cs-4 arm rose 10.5% from 0.83 to 0.93 L/min (p < 0.02) and ventilatory threshold 8.5% from 1.25 to 1.36 L/min. VO2max did not change in either group. No adverse events were reported (Chen et al., 2010). That paper is emerging as a small, within-group threshold signal in older, untrained adults, on a starch placebo, with a five-person dropout after randomisation, and with no VO2max effect. Citing it as a VO2max trial is a citation error. Hirsch and colleagues later cited it as 1 g per day (Hirsch et al., 2017). The capsule arithmetic in Chen's methods is 3 g. The dose is internally inconsistent in the source.

Hirsch, Smith-Ryan, Roelofs, Trexler, and Mock (2017) randomised 28 recreationally active adults to 4 g per day of a C. militaris-containing mushroom blend (PeakO2: oat-cultured mycelial biomass plus other species) or maltodextrin. After one week there was no time × treatment interaction for VO2max, ventilatory threshold, time to exhaustion, or anaerobic power; VO2max rose in both arms. Ten volunteers continued for two further weeks. In that exploratory cell, VO2max rose 4.8 ml/kg/min in the blend and 0.9 in placebo (interaction p = 0.042). Confidence intervals also favoured time to exhaustion in the blend (Hirsch et al., 2017). Grade: emerging for a 3-week signal in an n = 10 blend study; null at one week in n = 28; not a C. militaris isolate trial; not a trained-athlete trial.

Earnest and colleagues (2004) and Colson and colleagues (2005) tested a commercial formula whose primary listed ingredients were 1000 mg Cs-4 and 300 mg Rhodiola rosea, plus pyruvate, phosphates, ribose, adenosine, and chromium, in amateur cyclists. Earnest's n = 17, two-week loading-then-maintenance design found no treatment effect on peak VO2 (4.14 versus 4.10 L/min), time to exhaustion, peak power, or lactate thresholds (Earnest et al., 2004). Colson's n = 8 looked at muscle oxygen saturation (Colson et al., 2005). Those papers are strongly supported as null or uninformative for attributing any effect to Cordyceps, because the formula is not Cordyceps.

Yi, Xi-zhen, and Jia-shi (2004), cited by Hirsch as a 3 g per day, six-week elderly Cs-4 trial with VO2max +6.7%, is a Chinese Journal of Integrative Medicine paper that this build did not lock to a verified PubMed record. It is not used as a load-bearing number here.

The honest athletic summary is uncomfortable for both camps. Cordyceps is not a demonstrated ergogenic in trained cyclists at the Cs-4 3 g, five-week design Parcell ran. It is also not a null molecule in every older-adult threshold analysis. Stacking Parcell's null, Chen's threshold, and Hirsch's n = 10 blend into "Cordyceps raises VO2max" is the error.

11 Immune claims

Two randomised adult trials sit in this file, and they are not the same product.

Kang, Baik, Kim, Lee, Ahn, Park, and colleagues randomised healthy Korean men to 1.5 g per day of ethanol-treated C. militaris (n = 39) or microcrystalline cellulose/lactose placebo (n = 40) for four weeks. NK200, lymphocyte proliferation index, IL-2, and IFN-γ rose more from baseline in the C. militaris arm. The authors called the product safe and effective for enhancing cell-mediated immunity (Kang et al., 2015). That paper is emerging as a short biomarker trial in healthy men. It is not an infection-outcome trial. "Enhances immunity" as a consumer sentence is a grade inflation.

Jung and colleagues randomised healthy adults to 1.68 g per day of Paecilomyces hepiali CBG-CS-2 (n = 39) or control (n = 40) for eight weeks. NK cytotoxic activity rose 38.8% from baseline relative to placebo (p < 0.019). The title still says Cordyceps (Jung et al., 2019). Grade: emerging for that Paecilomyces preparation on an NK assay; not evidence about O. sinensis or C. militaris.

Das and colleagues' 2020 review of immune-stimulatory potentials is a secondary inventory (Das et al., 2020). Reviews do not create the missing infection trials.

12 Respiratory health

Wang, Li, Huang, Chen, and Chen randomised 120 people with moderate-to-severe persistent asthma to three months of Corbin capsule (C. sinensis formulation) plus inhaled corticosteroid and as-needed β-agonist, versus the same inhalers without Corbin. AQLQ scores, lung function, and several serum inflammatory markers moved in the add-on arm (Wang et al., 2016). The control arm did not receive a matched placebo capsule. That design is emerging as an open add-on signal and is not a blinded efficacy trial. Traditional "soothe the lung" language does not repair the missing placebo.

No locked Western trial shows that an oral Cordyceps supplement prevents respiratory infection or COPD progression. Grade for general respiratory-preventive use in healthy adults: speculative.

13 Metabolic outcomes

The metabolic file in this build is preclinical catalogues plus traditional hyperglycemia/hyperlipidemia lists (Zhu, Halpern, and Jones, 1998; Yue et al., 2013; Ashraf et al., 2020). Those are not a human metabolic-outcome programme. Li and colleagues' antioxidant-activity paper is a source-variation experiment, not an HbA1c trial (Li et al., 2001b). Grade for metabolic disease modification by oral Cordyceps supplements: speculative.

14 Kidney: traditional context and the Cochrane file

This is the one clinical domain with a Cochrane review, and the review is a warning.

Zhang, Lin, Tung, Kwan, Mok, Leung, and Chan included 22 Chinese studies, 1746 participants. Among people with CKD not on dialysis, Cordyceps preparations were associated with lower serum creatinine (14 studies, 987 people; MD −60.76 μmol/L, 95% CI −85.82 to −35.71), higher creatinine clearance (6 studies, 362 people; MD 9.22 mL/min), and less proteinuria (4 studies, 211 people; MD −0.15 g/24 h). Risk of bias was high in four studies and unclear in 18. The authors would not draw a definitive conclusion (Zhang et al., 2014). That review is strongly supported as a low-quality-evidence synthesis. It is not a licence to sell wild O. sinensis for kidney protection, and it pooled cultured mycelial products with the traditional name.

Li, Xue, Tian, Ding, Yan, Pan, and Feng randomised 202 renal-transplant recipients by lottery to Cordyceps 1.0 g three times daily plus usual immunosuppression (n = 93) versus usual immunosuppression (n = 109). Graft survival and creatinine did not differ. Uric acid and 24-hour protein were lower; reported hepatotoxicity and nephrotoxicity were lower; cyclosporine doses and troughs were lower at some months; chronic allograft nephropathy was 7.53% versus 18.35% (Li et al., 2009). Sun, Yang, Lu, Gao, Wang, Wang, and Tang compared Bailing capsule (a dry mycelial powder) against azathioprine in a 121-patient Chinese transplant series (Sun et al., 2004). Ding, Tian, Xue, and colleagues reported long-term Cordyceps with reduced cyclosporine (Ding et al., 2011). Those hospital papers are emerging as adjunctive Chinese-practice evidence. They are not Western confirmatory trials, and they are not supplement-aisle kidney tonics.

15 Aging

There is no locked human lifespan, healthspan, or aging-outcome trial of any Cordyceps preparation. Zhu, Halpern, and Jones listed "antisenescence" among review headings in 1998 (Zhu, Halpern, and Jones, 1998). Ashraf and colleagues listed "anti-aging" among cordycepin potentials in 2020 (Ashraf et al., 2020). Those headings are speculative as claims. An NK-cell movement in a four-week C. militaris trial is not aging (Kang et al., 2015). A mouse-liver ATP ratio is not aging (Dai et al., 2001). Grade for Cordyceps as a longevity agent: speculative.

16 Mycelium versus fruiting body, and what the grain is doing

Fruiting-body C. militaris, insect-complex wild O. sinensis, liquid-fermented mycelium, and solid-state mycelium on grain are different tissues on different substrates. Li, Yang, and Tsim treated that split as the reason analytical markers exist (Li, Yang, and Tsim, 2006). Hirsch's PeakO2 is explicitly "dehydrated, finely milled, cordyceps mycelial biomass powder cultured from organic oats" plus other species (Hirsch et al., 2017). Chen's Cs-4 was a Jiangxi GuoYao / Pharmanex fermentation powder filled into starch-matched capsules (Chen et al., 2010). Bailing is a dry mycelial powder used as a transplant adjunct (Sun et al., 2004).

A grain-grown mycelial biomass can meet a "Cordyceps powder" label while a large fraction of mass is leftover cereal. Ergosterol and nucleoside methods exist to push back on that (Yang et al., 2009; Fan et al., 2008). They are not routinely the consumer-facing specification. Grade of "mycelium on oat equals fruiting body": speculative, and rejected.

17 Substitution, adulteration, and standardisation

Li, Yang, and Tsim named the market problem in one paragraph: rarity of the wild complex, fermented mycelia sold as health food, C. militaris used as a substitute, adulterants confusing the market, therefore quality control (Li, Yang, and Tsim, 2006). Lo and colleagues' mysterious-caterpillar review is the same warning in a later key (Lo et al., 2013). Chen and colleagues' anamorph paper is why Paecilomyces products are a substitution, not a technicality (Chen et al., 2001).

Standardisation to adenosine, cordycepin, or "7% cordyceps polysaccharides" does not close that file. Those assays can be passed by the wrong species, a spike, or a polysaccharide-rich grain fraction. Authentication that would actually bite is DNA (ITS as in Chen et al., 2001) plus chemistry plus a declaration of tissue and substrate. Commercial certificates rarely do all three.

18 Safety, at the same volume as efficacy

LiverTox's 2025 chapter is the load-bearing safety synthesis in this build. Cordyceps extracts have not been associated with aminotransferase elevations or clinically apparent acute liver injury. Likelihood score E. Adverse events in trials were described as uncommon: abdominal discomfort, diarrhoea, dry mouth, nausea, poor appetite, rash. A single case of hepatoportal sclerosis in an elderly woman improved after stopping a long-term cordyceps product. LiverTox also writes that FDA designates cordyceps as GRAS (LiverTox, 2025). That GRAS sentence is LiverTox's. This article did not independently verify a GRAS notice and does not treat it as a safety proof of aisle products, whose identity is the problem in sections 16–17.

Jacobsson, Jönsson, Gerdén, and Hägg's Swedish spontaneous-report series is cited by LiverTox as containing no cordyceps attributions among 778 herbal reports (LiverTox, 2025). Absence in one registry is not a global safety trial.

Kang's four-week C. militaris trial and Jung's eight-week Paecilomyces trial reported no statistically significant adverse-reaction excess (Kang et al., 2015; Jung et al., 2019). Short healthy-adult biomarker trials are emerging as short-term tolerability notes. They are not a chronic-use or contaminated-wild-product file.

Wild material carries a separate, plausible contamination and heavy-metal risk that this build did not lock to a single clean analytical series; Li, Yang, and Tsim's QC review exists because adulteration is not theoretical (Li, Yang, and Tsim, 2006). Transplant and CKD papers are hospital adjuncts, not over-the-counter safety proofs (Zhang et al., 2014; Li et al., 2009).

This document specifies no dose. Every milligram figure above is a reported study parameter attached to a named product and population.

19 Five research matrices

Species and formulation

Exercise

Clinical (non-exercise)

Chemical analysis

Safety

20 Critical questions

Does traditional use licence modern exercise, immune, kidney, or aging claims? No. Zhu, Halpern, and Jones collected a traditional and early-clinical catalogue (Zhu, Halpern, and Jones, 1998). Parcell, Chen, Hirsch, Kang, Jung, Wang, and Cochrane each tested a different object. The traditional kidney/lung sentence is not Parcell's cyclists and is not Kang's NK200. Grade of the extrapolation: speculative, and rejected.

Is species substitution a footnote? No. Sung and colleagues split the genus (Sung et al., 2007). Chen and colleagues showed the wild anamorph is Hirsutella, not Paecilomyces (Chen et al., 2001). Jung's title says Cordyceps and the methods say Paecilomyces hepiali (Jung et al., 2019). Hirsch's title says C. militaris and the product is a multi-species oat mycelium blend (Hirsch et al., 2017). Substitution is the manufacturing system.

Are the exercise trials large enough to carry a VO2max claim? No. The only trained-endurance Cs-4 trial in this file is a null in 22 men (Parcell et al., 2004). Chen's VO2max did not move (Chen et al., 2010). Hirsch's VO2max interaction appears in ten people on a blend (Hirsch et al., 2017). Earnest's formula is not Cordyceps (Earnest et al., 2004). Tiny positive cells do not outrank a trained null by being newer or more photogenic.

Does rodent mitochondrial evidence show that Cordyceps "boosts ATP"? No. Dai, Bao, Xu, Cooper, and Zhu measured mouse-liver 31P NMR after seven days of CordyMax: β-ATP +12.3% and +18.4% at 200 and 400 mg/kg, Pi down, ratio up, reversible after washout (Dai et al., 2001). Manabe and colleagues had already reported a higher hepatic ATP/Pi ratio in mice given cultured C. sinensis mycelial extract (Manabe et al., 1996). Those papers are strongly supported as murine hepatic NMR. Dai and colleagues then wrote that the finding suggested a mechanism "underlying the known clinical effectiveness of CordyMax in alleviating fatigue and improving physical endurance, especially in elderly subjects" (Dai et al., 2001). That sentence is the slogan. It is not a human 31P-MRS muscle study, and "known clinical effectiveness" was not known then and is not known now at the athletic endpoint Parcell tested.

Do product-identity problems make the rest of the file unreadable? They make pooling unreadable. They do not make every paper unread. Parcell is a Cs-4 paper. Chen is a Cs-4 paper. Kang is a C. militaris ethanol-extract paper. Jung is a Paecilomyces paper. Cochrane is a Chinese-preparation pool. Read that way, the file is small and split. Read as "Cordyceps," it becomes a legend.

21 What is known, and what is sold

What is known is sharp. The wild prestige object is Ophiocordyceps sinensis, anamorph Hirsutella sinensis, not a Paecilomyces fermentation and not C. militaris (Sung et al., 2007; Chen et al., 2001). Cordycepin is a C. militaris-associated 3'-deoxyadenosine, not a synonym of the wild complex (PubChem, 2026a; Ashraf et al., 2020). In trained cyclists, Cs-4 at 3 g per day for five weeks did not improve endurance (Parcell et al., 2004). In older adults, Cs-4 moved a metabolic threshold and did not move VO2max (Chen et al., 2010). A mushroom blend moved VO2max in ten people after three weeks and not in twenty-eight after one (Hirsch et al., 2017). Cochrane would not conclude on CKD (Zhang et al., 2014). LiverTox would not pin acute liver injury on the extracts it reviewed (LiverTox, 2025). Mouse-liver ATP is mouse-liver ATP (Dai et al., 2001; Manabe et al., 1996).

What is sold is a collapse. One word, a caterpillar photograph, a cordycepin percentage, an ATP slogan, and a VO2max implication. The chemistry is real. The wild fungus is real. The slogan is the collapse.

A reader who wants a single sentence can have it. Cordyceps is established as a split taxonomic and commercial object; it is strongly supported as a null endurance aid in the only trained Cs-4 cycling trial locked here; it is emerging as a small older-adult threshold signal, as NK-cell biomarker movement for specific C. militaris or Paecilomyces preparations, and as low-quality Chinese CKD adjunct evidence; it is not a demonstrated longevity agent; and "ATP boosting" remains a murine hepatic NMR finding asked to do human work it has not done (Parcell et al., 2004; Chen et al., 2010; Kang et al., 2015; Jung et al., 2019; Zhang et al., 2014; Dai et al., 2001).

The extra length is the point. Split back into its parts — the polysaccharide chemistry, the cultured Cs-4 mycelium, the cordycepin that does not survive in blood, and the two small human ergometry trials — the word names a supplement with a real pharmacology and no demonstrated effect on human performance at the doses sold.

22 References

23 Evidence handling

Findings are labelled in the reporting sentence by design: phylogenetic revision, anamorph sequencing, randomised trial, Cochrane review, analytical chemistry, rodent NMR, regulator/database record, or labelled secondary reporting. Animal and in-vitro results are used here only where they are the load-bearing source of a slogan (ATP) or a chemical inventory. They are not used to imply a human outcome.

Conflicting evidence is kept in the same section. Parcell's null is not averaged with Hirsch's n = 10 cell. Chen's threshold is not rewritten as a VO2max effect. Jung's Paecilomyces product is not renamed O. sinensis because the title did so. Cochrane's creatinine mean difference is not promoted past the authors' refusal to conclude.

Quantitative claims were locked to verified NCBI records, to PMC full text of Chen 2010, Hirsch 2017, Sung 2007, and Zhang 2014, to PubChem compound records, and to the LiverTox chapter fetched for this build. Candidate PubMed identifiers that resolved to a different paper were discarded and not cited.

References are generated from those verified records plus the non-PubMed database and LiverTox sources listed with them. Internal production logs are not part of this apparatus.