Corydalis

What is Corydalis?

A 2025 analysis of 14 Corydalis yanhusuo supplements sold online in the United States found total alkaloid content ranging from below the limit of quantification to 11.4 mg per gram, so the richest product carried more than ten times the alkaloid of the poorest quantifiable one, two carried nothing measurable and a third almost nothing. That finding, from Luis and Schneider at Vanderbilt (Frontiers in Pharmacology, January 2025), is the frame for everything below.

Corydalis yanhusuo W.T. Wang is a small perennial in the poppy family (Papaveraceae) cultivated mainly in Zhejiang, Jiangxi and Anhui provinces of China. The medicinal part is the dried tuber, boiled or steamed until the white core disappears and then sun-dried. The Chinese Pharmacopoeia lists it under the name yanhusuo (Corydalis Rhizoma), requires the crude drug to contain not less than 0.050% tetrahydropalmatine on a dried basis, and lists a decoction dose of 3 to 10 g for practitioner use. The tuber has been in the Chinese materia medica since at least the Tang dynasty (618 to 907), first recorded for chest pain and later for postpartum and abdominal pain.

In the US it is sold as a dietary supplement: whole-tuber powder, 10:1 or 20:1 water extracts in 400 mg to 1,000 mg capsules, and alcohol tinctures. Its pharmacology rests on isoquinoline alkaloids of the protoberberine and tetrahydroprotoberberine classes, of which more than 80 have been isolated. The tuber contains roughly 0.5% to 1.3% total alkaloids by weight depending on origin, processing and assay.

Common Names

Corydalis, yan hu suo, yanhusuo, yuanhu, Corydalis Rhizoma, Corydalis tuber, Rhizoma Corydalis, "corydalis extract", Chinese poppy (a marketing name, not a botanical one).

Primary Active Compounds

Dehydrocorybulbine (DHCB), the compound in the University of California Irvine analgesia work; levo-tetrahydropalmatine (THP), a dopamine D1 and D2 receptor antagonist sold in China as the prescription drug Rotundine; dehydrocorydaline, the most abundant alkaloid by mass in good material; corydaline; protopine; tetrahydroberberine; coptisine; palmatine; columbamine; glaucine. Corydalis also contains protoberberine alkaloids, berberine itself among them, but that compound is covered in the Coptis, Goldenseal, Barberry, Oregon Grape and Yellow Root issues and is not this story.

Key Note

The alkaloid that made headlines as a "new painkiller from Chinese medicine" was injected into mice at 5 to 40 mg per kilogram. One gram of the richest US product tested held about 0.7 mg of it.

This issue is about the distance between the molecule in the paper and the powder in the capsule, and about the second alkaloid, THP, which carries a regulatory and case-report history that no product label mentions.

Primary Functions & Benefits

Pain, in rodents, with a clear mechanism

The mechanism belongs to this issue, so here it is in full. The isoquinoline alkaloids of Corydalis do not work through opioid receptors in any meaningful way. In the 2014 Current Biology study, DHCB was a weak partial agonist at the human mu opioid receptor (half-maximal response at 100 micromolar in HEK293 cells), and its antinociception in mice was not blocked by naloxone at 1 mg per kilogram. Its affinities at the five dopamine receptors were in the submicromolar to micromolar range, more than a hundred times higher than at mu, with the highest at D2. A selective D2 agonist (quinpirole, 0.5 mg per kilogram) reversed DHCB antinociception; a D1 agonist (SKF-38393) did not; and in dopamine D2 receptor knockout mice, 5 mg per kilogram of DHCB produced no significant tail-flick effect. The 2016 PLoS One extract study (Wang, Zhang, Civelli and colleagues) screened the whole water extract against 20 G protein-coupled receptors and found antagonist activity at only two, dopamine D1 and D2, each with an IC50 of about 3.1 micrograms per milliliter. In D2 knockout mice the extract lost its effect on acute thermal pain and on mechanical allodynia after spinal nerve ligation, but kept its effect in the formalin inflammatory test, so the inflammatory component runs through something else.

THP works along the same axis. Wang and Mantsch (Future Medicinal Chemistry, 2012) give its Ki at D1 as about 124 nM and at D2 as about 388 nM, with weaker binding at D3 (1.4 micromolar), significant binding at 5-HT1A (Ki about 340 nM) and alpha-2 adrenergic receptors, antagonism at alpha-1 adrenergic receptors, plus positive allosteric effects at GABA-A. Dopamine antagonists are not normally thought of as painkillers, and the proposed explanation is that at low doses these compounds preferentially block presynaptic D2 autoreceptors, which raises dopamine release in pain-modulating circuits. The same mechanism is why THP sedates.

Pain, in humans

One controlled trial exists. Yuan and colleagues at the University of Chicago (Journal of Clinical Pharmacology, 2004) gave adult volunteers a single oral dose of extracts of Corydalis yanhusuo and Angelica dahurica, the pairing used in the Chinese patent formula Yuanhu Zhitong, in a double-blind crossover design, and measured response to the cold pressor test. Pain intensity and pain bothersomeness scores both fell (P less than 0.01) with a dose-related effect. I could not obtain the full paper, so the participant number, the exact doses and the preparation are not stated here, and the result is for a two-herb combination, not Corydalis alone.

Addiction research, on THP, not on the herb

THP has been tested in humans for cocaine and heroin use. Hassan and colleagues (Journal of Clinical Pharmacology, published online 2016) gave 24 cocaine-using men 30 mg of purified levo-THP twice daily or placebo for 4 days; 20 completed, and THP did not change cocaine pharmacokinetics or acute cardiovascular response. That is a safety and pharmacokinetic study of a purified drug, and none of it transfers to a capsule of tuber powder.

Cardiovascular and other claims

Dehydrocorydaline and the other quaternary alkaloids protected cultured heart cells in a hypoxia-reoxygenation assay across 37 sourced samples (Li and colleagues, Journal of Ethnopharmacology, 2017). That is cell culture. There are no human cardiovascular outcome data for the herb.

Forms & Standardization

Whole tuber powder

The Chinese Pharmacopoeia crude drug: tubers boiled or steamed until the white core is gone, then dried, sliced or powdered. The 2020 edition specifies not less than 0.050% THP for the crude drug and not less than 0.040% for prepared slices and vinegar-processed material, on a dried basis. The Hong Kong Chinese Materia Medica Standards monograph sets a different marker: not less than 0.10% for corydaline plus THP combined. These are minimums, not targets, and they say nothing about DHCB or dehydrocorydaline. A 2021 review summarizing HPLC surveys of 15 Chinese production areas reported corydaline at 0.08% to 0.44% and THP at 0.03% to 0.2% across origins, so a tuber can miss the pharmacopoeial floor by a factor of two simply by where it grew.

Vinegar-processed tuber (cu yanhusuo)

The Pharmacopoeia's standard processing: rice vinegar at 20 mL per 100 g of sliced tuber, absorbed, then stir-fried over low heat until dry. The claim is that vinegar converts tertiary alkaloids to more soluble salts. The measured effect is modest. In Molecules (2014), a Beijing and Vienna group (Wu, Kopp and colleagues) quantified 11 alkaloids and found that in a water decoction, vinegar processing raised THP from 0.321 to 0.381 mg per gram (plus 18.7%) and corydaline from 0.189 to 0.233 mg per gram, while the methanol-extractable total fell 6.1%. Dehydrocorydaline, a quaternary alkaloid, fell 9.7% in the methanol extract with vinegar and 22.9% with wine. The UC Irvine extract was made from vinegar-processed tuber.

Water extracts labeled by ratio

Most US capsules say "10:1" or "20:1" and nothing else. The 2016 UC Irvine extract is a useful reference for what those ratios mean: 10 kg of ground, vinegar-processed tuber decocted twice in 100 L of water (120 minutes then 90 minutes), pooled and spray-dried to 0.6 kg, about 17:1 by weight, containing roughly 0.25% THP and 0.18% DHCB. Nobody selling a 20:1 extract tells you whether the second decoction was done, how long it ran, or what the extract assayed at. In the Vanderbilt survey, the five best products (9.5 plus or minus 1.6 mg per gram total alkaloids) sat below the ground tuber reference (12.7 mg per gram), so "extract" on a US label mostly means "powder that has been through hot water," not a concentrate. One product labeled 20:1 (sample 3, 500 mg capsules) assayed at 3.5 mg per gram.

Alcohol tinctures

Two of the 14 Vanderbilt products were liquids. One label stated a dry herb to menstruum ratio of 1:3 and 233 mg herb per 0.7 mL. Alkaloid content per serving in liquids was at the low end of the range.

"Corydalis extract" with no species

Two products whose Supplement Facts panel said only "Corydalis" (samples 8 and 10; sample 8 did say yan hu suo elsewhere on the pack) contained no quantifiable isoquinoline alkaloids at all, and a fourfold larger sample extraction gave the same result. The genus has several hundred species. A label that says "Corydalis" without "yanhusuo" is not making a claim you can hold it to.

Purified THP

A licensed prescription drug in China with no US drug approval, and I could find no US product lawfully marketing it as a dietary ingredient; the near-pure THP tablet product of the early 1990s (Jin Bu Huan, below) is the reason it is treated as a drug rather than a supplement constituent. One of the 14 surveyed supplements nonetheless assayed at 5,217 ng per mg THP, more than ten times the next highest, with the rest of the alkaloid profile nearly absent; a repurchased bottle of the same product assayed at 110 ng per mg. The authors concluded the first batch had been spiked with the pure compound.

Which form the trials used

The mouse work used synthetic DHCB (2014) and a vinegar-processed water extract (2016), both injected. The one human trial used an oral two-herb extract. No trial has used a US retail capsule.

Food Sources

Corydalis is not a food. No part of the plant is eaten, and the alkaloids that define it (isoquinolines in the protoberberine and tetrahydroprotoberberine series) do not occur in any common food. Dietary intake is zero. Exposure comes from four routes: Chinese decoctions prescribed at 3 to 10 g of tuber per day by practitioners; Chinese patent medicines such as Yuanhu Zhitong tablets (Corydalis plus Angelica dahurica); US capsules and tinctures; and, historically, THP-containing tablets sold under other botanical names, which is the subject of the next three sections.

Who Should Take Corydalis

The honest answer is that no group has been shown to benefit from a US retail Corydalis product in a published trial. The people who buy it are adults with musculoskeletal pain, menstrual pain, nerve pain or insomnia who want something non-opioid. For them the relevant facts are these.

The pharmacology is real and the dose-response in animals is clear, but the effective doses were 100 to 500 mg per kilogram of water extract injected into mice, and 1 mg per kilogram of its two named alkaloids did nothing. Whether a 1,000 mg capsule, taken orally by a 70 kg adult, delivers enough of anything to reproduce the mouse result has not been tested.

A reader who decides to try it anyway should buy a product that names Corydalis yanhusuo on the Supplement Facts panel, states the plant part (tuber or rhizome), and comes from a maker who will produce a certificate of analysis with a named alkaloid value. In the 2025 survey, two of the three products labeled only "Corydalis" with no species, and the one product that named DHCB on the label without stating an amount, were the ones that assayed near zero; the exception, a plain "Corydalis poppy powder" with no Supplement Facts panel at all, was among the richest, so the label is a screen, not a guarantee in either direction.

Someone with chronic daily pain who takes this instead of seeing a clinician is making a worse decision than someone who takes it alongside one. The dopamine receptor pharmacology means it is not pharmacologically inert, and the sections on contraindications and interactions below are not optional reading.

Who Should AVOID or Use Caution

Contraindications

• Pregnancy. Traditional Chinese sources caution against Corydalis in pregnancy and in postpartum blood deficiency with bleeding, and the 2020 review by Zhang and colleagues in Natural Product Communications repeats that caution. There are no reproductive toxicity data for the extract. A dopamine antagonist with no pregnancy data is a contraindication, not a caution.

• Breastfeeding. Same reasoning. THP crosses into brain tissue readily in rats; there are no milk transfer data in humans.

• Children. The 1993 Colorado cases were children aged 13 months to two and a half years who ate THP tablets and within an hour developed lethargy and respiratory depression, with bradycardia in the two larger ingestions, one to 30 to 35 beats per minute. That product was not Corydalis (see Toxicity Symptoms), but the alkaloid is the same one the Chinese Pharmacopoeia uses as Corydalis's marker. Keep every Corydalis product out of reach and do not give it to a child.

• Parkinson's disease or anyone taking levodopa, pramipexole, ropinirole or rotigotine. THP and DHCB are dopamine D1 and D2 antagonists. Blocking the receptors those drugs act on is a direct pharmacological opposition.

• Known liver disease. The 1993 MMWR editorial note on the Los Angeles hepatitis cases stated that people with known liver disease should avoid the THP product involved, and a 2025 analysis of China's national adverse drug reaction database (Guo and colleagues, China Journal of Chinese Materia Medica) flagged Corydalis Rhizoma itself as a high-signal herb for liver injury.

Use Caution

• Anyone on a sedative, hypnotic, benzodiazepine, opioid, gabapentinoid or alcohol. THP sedates; the UC Irvine group found their extract reduced locomotor activity in Swiss Webster mice at doses above 200 mg per kilogram, and set 200 mg per kilogram as the non-sedating ceiling for the rest of their work.

• Anyone on an antipsychotic (haloperidol, risperidone, aripiprazole, quetiapine, olanzapine). Additive dopamine blockade is a plausible route to extrapyramidal symptoms; Wang and Mantsch note that THP overdose produces extrapyramidal signs.

• Anyone on a drug cleared by CYP3A4 with a narrow margin: tacrolimus, cyclosporine, certain statins. In beagle dogs, THP at 12 mg per kilogram three times daily raised midazolam exposure by 39% to 57% (Zhao and colleagues, Phytotherapy Research, 2016). Human data do not exist.

• Anyone with bradycardia, heart block or on a beta blocker, digoxin, diltiazem or verapamil, because the pediatric THP overdoses presented with life-threatening bradycardia.

• Drivers and machine operators, for the sedation.

Critical Safety Point

The best-known hepatitis cases are tied to a THP-containing tablet product whose botanical identity was never confirmed as Corydalis and was most likely Stephania. That does not clear Corydalis. The alkaloid that product delivered at 28.8 mg per tablet is the same alkaloid that is the Pharmacopoeia marker for Corydalis, Chinese pharmacovigilance now flags Corydalis Rhizoma for liver injury, and one US supplement in 2025 assayed at 5 mg of THP per serving. More directly, a 2023 US case series (Engman and colleagues, ACG Case Reports Journal) described a 51-year-old woman and a 52-year-old man who developed hepatitis on a sleep aid sold at cannabis dispensaries that listed corydalis as an active ingredient; the woman's biopsy showed severe acute hepatitis and her enzymes rose again when she restarted the product. The product was not assayed and the species was not stated, so it cannot be tied to Corydalis yanhusuo, but it is a US report on a retail corydalis product rather than a THP tablet. If you develop fatigue, nausea, dark urine, pale stools, itching or yellowing of the eyes while taking any Corydalis product, stop it and get liver enzymes measured. Two of the seven patients in the 1994 Annals of Internal Medicine series relapsed abruptly on rechallenge, and so did the 2023 woman.

Recommended Dosages

There is no clinically established dose for a Corydalis supplement, because no published trial has dosed one. What follows describes what is sold and what it contains, not what works. Rodent doses and Chinese prescription doses are deliberately kept out of this section and appear, scoped, in Special Considerations.

Whole tuber powder capsules

Sold at 400 mg to 1,000 mg per capsule, with labels suggesting one to three capsules daily. At the 2025 survey's tuber reference value of 12.7 mg total alkaloids per gram, 1,000 mg of good powder delivers about 12.7 mg of mixed alkaloids, of which roughly 2.8 mg is dehydrocorydaline, 0.85 mg is DHCB and 0.6 mg is THP. At the survey's low group (1.8 mg per gram) the same capsule delivers 1.8 mg, and from the three near-empty products, essentially nothing.

Water extract capsules (10:1, 20:1)

Sold at 500 mg to 1,000 mg per capsule. In the survey the ratio on the label did not predict content: the best products assayed at 7.5 to 11.4 mg per gram, no better than raw tuber, and one 20:1 product assayed at 3.5 mg per gram. Treat the ratio as unverified marketing unless the certificate of analysis gives an alkaloid value.

Tinctures

Labels suggest 0.7 mL to 5 drops per serving. The two liquids in the survey assayed at 0.9 and 2.0 mg per gram.

What the THP content means in practice

A 1,000 mg capsule of the richest tuber powder delivers under 1 mg of THP, a small fraction of the daily amount in the Chinese prescription drug (that figure is given, scoped, in Special Considerations). That is reassuring for the sedation and liver questions and unhelpful for the pain question. The adulterated product in the survey delivered 5 mg per serving, which at three servings a day is 15 mg, about a quarter of the lowest prescription dose of the pure drug.

Duration

No duration has been studied for any oral product in humans. The hepatitis cases on the THP tablet product appeared after weeks to months of regular use (details in Toxicity Symptoms). If a reader uses a Corydalis product at all, a sensible pattern is intermittent use for a defined pain episode rather than open-ended daily use, with liver enzymes checked if use runs beyond a few weeks.

Timing & Administration

Nothing in the human literature addresses timing. The Chinese Pharmacopoeia gives the powder as a per-dose amount rather than a daily one, implying divided dosing, and purified THP is prescribed in China for insomnia as well as pain, which tells you where the sedation lands. For a reader using a capsule for daytime pain, that argues for a single trial dose in the evening first, to learn whether the product sedates them, before taking it during the day. Alkaloid salts dissolve in an acidic stomach, so taking it with food is not a problem. Alcohol adds to sedation.

Timeline of Effects

Every timing number here is from mice or from the purified drug, labeled as such; none is a schedule for taking a capsule.

Onset

In the mouse studies, injected DHCB and injected extract produced measurable tail-flick effects at 30 minutes, the first time point tested. Oral kinetics are slower. In rats given THP by mouth, blood levels peak within the first hour and fall with a half-life of about 5 hours; in humans the half-life of levo-THP is reported at about 10 hours (Wang and Mantsch, 2012). The one human trial (Yuan, 2004) measured cold pressor response after a single oral dose, so any effect it detected was within hours. No oral human timing data exist for DHCB.

Peak and duration

In the 2014 mouse study, DHCB remained in plasma at relatively high concentrations for at least 3 hours and the tail-flick effect lasted at least 3 hours at the higher injected doses. The 2016 extract study found effects in neuropathic pain assays for at least 2 hours after a non-sedating injected dose.

Fair trial and when to stop

There is no trial-derived duration. For a pain supplement with acute-onset pharmacology the honest trial is days, not months: if a product gives no perceptible change in a defined pain within three or four doses, there is no evidence that continuing will change that. The mouse tolerance data argue against escalating doses; they do not argue for long courses.

Benefits of Taking Corydalis

The benefit case rests on the mechanism and the animal data above, plus one human cold pressor trial of a two-herb combination. Set against that is the content survey. A product that actually contains the alkaloids will deliver a few milligrams of mixed dopamine-antagonist isoquinolines per capsule; whether that reduces pain in a person has not been tested. What the survey does give a buyer is the label features that predicted a real product: "Corydalis yanhusuo" by species, "tuber" or "rhizome" by part, and a maker who would state alkaloid content.

Potential Negatives & Side Effects

Sedation and dizziness

THP is approved in China partly as a hypnotic, and the Chinese labeling lists sleepiness, dizziness and nausea as the adverse effects above the therapeutic range (Wang and Mantsch, 2012). The UC Irvine extract reduced locomotion in mice above 200 mg per kilogram. A 1984 Chinese clinical observation cited in the Hong Kong Polytechnic University herb database reports drowsiness, dizziness and abdominal distension with larger doses of the herb. At the sub-milligram THP content of a legitimate capsule, overt sedation is unlikely; with a spiked product it is not.

Liver

The best-documented cases are on a THP tablet product, detailed under Toxicity Symptoms. There are also case reports on corydalis products themselves: the 2023 US series of two adults on a corydalis-containing dispensary sleep aid (above, under Critical Safety Point), a 2021 Chinese report in Clinical Toxicology of repeated liver injury on rechallenge with Kedaling tablets, a Chinese drug made from the total alkaloids of Corydalis saxicola, and a 2009 Korean report of cholestatic hepatitis that recurred when a man unknowingly re-ate Corydalis speciosa. None of these involved a measured Corydalis yanhusuo product, and together they are a handful of cases against decades of use, but they are enough to say the signal is not confined to Jin Bu Huan. A 2021 review of processing in Evidence-Based Complementary and Alternative Medicine summarized Chinese animal data in which raw Corydalis extract above the equivalent of 1 to 3 g per day, or total alkaloids at 473 mg per kilogram, produced liver damage, muscle tremor and renal hemorrhage in mice. The 2025 Chinese pharmacovigilance analysis flagged Corydalis Rhizoma as high-risk for liver injury and named THP and tetrahydroberberine as the suspected components.

Extrapyramidal effects

Reported by Wang and Mantsch for THP overdose, consistent with any D2 antagonist. Not reported for the herb at supplement doses.

Bradycardia and respiratory depression

Documented only in the pediatric THP tablet overdoses, at 7 to 60 tablets of 28.8 mg each. Not reported with the herb.

Allergic reactions

Wang and Mantsch report a few cases of allergic reactions to Rotundine. The second Los Angeles hepatitis patient had a biopsy showing eosinophils in the portal tracts, which is the histology of a drug hypersensitivity reaction rather than direct toxicity.

Gastrointestinal

Nausea is listed for THP; abdominal distension for the herb (above).

Deficiency Symptoms

What Corydalis addresses

No deficiency state exists. Isoquinoline alkaloids are plant defense compounds with no role in human physiology, and a person who has never consumed Corydalis is missing nothing. What the herb is used for, pain and sleeplessness, are symptoms, not deficiencies, and the pharmacology that addresses them (dopamine receptor antagonism) is the pharmacology of a drug, not a nutrient.

Bottom line

There is no deficiency to correct, no biomarker to measure and no baseline to restore. Compare magnesium, where a serum or red cell value can tell you something; nothing comparable exists here.

Toxicity Symptoms

At high intake

The human toxicity record is on purified THP, and the preparation matters, so here is the record with the preparation named each time.

Jin Bu Huan Anodyne Tablets, sold in US health food stores in the early 1990s, were labeled as containing a "Polygala chinensis" alkaloid at 30% with 70% starch on some packaging (MMWR, December 3, 1993) and as "Lycopodium serratum" in the Annals of Internal Medicine series (Woolf and colleagues, November 1994). Analysis at Colorado State University found the tablets were 36% levo-THP by weight, 28.8 mg per tablet, with no other plant alkaloid detected. Neither Polygala nor Lycopodium makes THP. The MMWR authors noted that THP occurs in Stephania and Corydalis. The Hong Kong Hospital Authority's poisonous plant register lists "Jin Bu Huan" as a Chinese name for Stephania sinica, a Menispermaceae species whose root is used medicinally and whose toxic principle is THP. The botanical source of the US tablets was never confirmed in the published record; what can be said is that they were a near-pure alkaloid pressed into tablets, not a Corydalis yanhusuo preparation in any form a supplement buyer would recognize.

Three Colorado children (13 months, 23 months and two and a half years) ate about 60, 7 and 17 tablets respectively in 1993. All three were found within an hour and were lethargic; the two larger ingestions produced respiratory depression, one with a pulse of 30 to 35 that needed atropine and intubation, the other with a transient pulse in the mid 40s. Naloxone did nothing in either of the two tested, consistent with the non-opioid pharmacology. All recovered fully, within hours to a day.

Three Los Angeles women aged 24, 45 and 66, taking 2 to 4 tablets at night two to four times a week for insomnia or back pain, developed acute hepatitis within 1 to 3 months in 1993, with ALT values of 786, 1,308 and 1,468 U/L and one total bilirubin of 28.0 mg/dL. All bought the product at the same store. One resumed the tablets after recovering and her ALT climbed back to 961 U/L within weeks. Woolf and colleagues then assembled seven cases: hepatitis after a mean of 20 weeks (range 7 to 52) of use, resolving in six of the seven within a mean of 8 weeks (range 2 to 30) after stopping, with recrudescence in two who restarted. Picciotto and colleagues (Journal of Hepatology, 1998) reported a 49-year-old man on 3 tablets daily who developed biopsy-proven chronic hepatitis with moderate fibrosis within 2 months, and McRae and colleagues (European Journal of Gastroenterology and Hepatology, 2002) found Jin Bu Huan implicated in 11 published patients.

In Hong Kong, Lai and Chan (Clinical Chemistry, 1999) diagnosed nine adult THP overdoses in 16 months from "proprietary biopharmaceuticals containing purified THP," with serum THP of less than 0.1 to 1.2 mg per liter in the five measured. All recovered quickly after mild neurological disturbance. Luis and Schneider, citing that paper, state that two of those patients were assumed to have taken up to 1,500 and 2,000 mg of THP.

Signs to reduce or stop

Drowsiness beyond what you expect, slowed pulse, nausea, right upper abdominal discomfort, dark urine, itching or jaundice. Stop immediately for any of the liver signs and get ALT, AST, alkaline phosphatase and bilirubin measured. If a product is making you sleepy at label dose, the content is either higher than a legitimate tuber product can deliver or you are unusually sensitive; either way, stop it.

General note

I could find no published case that ties a measured, species-confirmed Corydalis yanhusuo supplement to liver injury, but there are now cases on corydalis products whose species and content were not established: the two 2023 US dispensary sleep-aid patients, the Chinese Kedaling (Corydalis saxicola) rechallenge report and the Korean Corydalis speciosa case, all described under Potential Negatives. The reasons for caution are those cases, that the toxic alkaloid in the Jin Bu Huan record is Corydalis's own marker compound, that a US product was found spiked with it in 2025, and that China's own adverse reaction database now flags the herb. The acetic acid extract (total alkaloids) of the tuber has an oral LD50 of 0.86 g per kilogram in mice, with most deaths within 3 hours and renal arteriole hemorrhage on microscopy (cited in Zhang and colleagues, 2020). That is a mouse value for a concentrated alkaloid fraction and should not be scaled to a capsule, but it is not a benign number.

How Corydalis Works

The receptor-level mechanism is in Primary Functions and is not repeated here. Four further points explain why the herb and its isolated alkaloids behave differently.

Stereochemistry

THP has two enantiomers. The levo form is the dopamine antagonist and the analgesic; the dextro form has distinct actions including monoamine depletion and, per Wang and Mantsch, may contribute to the toxicity of racemic preparations. Plant material is predominantly the levo form, but Ma and colleagues (Molecules, 2008, from McLean Hospital and Harvard) found it necessary to measure the d to l ratio of each tertiary alkaloid isolated from Corydalis yanhusuo by chiral HPLC, and a synthetic THP spike could be fully racemic. No supplement label states the enantiomer.

The whole extract outperforms its named parts

In the 2016 study, 500 mg per kilogram of extract (about 1 mg per kilogram each of THP and DHCB) raised tail-flick latency significantly while the 1 plus 1 mg per kilogram mixture of the two pure compounds did not. Either other alkaloids contribute (the authors name palmatine, columbamine and glaucine as candidates with rodent antinociceptive data), or the matrix changes absorption, or both. This is the strongest argument for using the whole tuber rather than a "standardized" extract that chases one marker.

Multiple targets beyond dopamine

Independent groups have reported that THP acts at the spinal sigma-1 receptor (Kang and colleagues, Scientific Reports, 2016) and that Corydalis alkaloids block voltage-gated sodium channels, and corydaline and corydine have been characterized as G protein-biased mu opioid agonists in mice after subcutaneous dosing. Dehydrocorydaline, the most abundant alkaloid, reduced acetic acid writhing and formalin licking in mice at 3.6 to 10 mg per kilogram intraperitoneally, and that effect was partly naloxone-sensitive, so the dominant alkaloid may run through a different pathway from the two famous ones.

Absorption and processing

Alkaloids are bases; as salts they dissolve in water, which is why decoction works and why vinegar processing modestly raised decoction THP. In rats, processed extract prolonged the elimination half-life of THP, dehydrocorydaline and protopine, with mean residence time over 6 hours. THP reaches peak brain levels within the first hour in rats. DHCB is not metabolized by phase I enzymes and is slowly glucuronidated, which is why it persisted in mouse plasma for 3 hours. In beagle dogs, the Yuanhu Zhitong combination raised the Cmax and AUC of dehydrocorydaline, THP, protopine, palmatine and columbamine compared with raw tuber alone.

Synergistic Supplements

Angelica dahurica (bai zhi)

The only pairing with human data. The Yuanhu Zhitong formula is Corydalis with Angelica dahurica, and the 2004 Chicago cold pressor trial used the two together. Dog pharmacokinetics show the combination raises exposure to the Corydalis alkaloids. If a reader is going to use Corydalis for pain, this is the combination with the best evidence base, thin as it is.

Melia toosendan (chuan lian zi)

Jin Ling Zi San is Corydalis with Melia toosendan fruit, a classical pairing for abdominal pain. In the 2014 Molecules study, co-decocting the two pulled more alkaloid into the water than Corydalis alone, and wine-processed tuber raised every tertiary alkaloid measured in that decoction. Melia toosendan contains toosendanin, which has its own hepatotoxicity record in animals, so this is not a pairing to assemble from supplement bottles.

What not to stack

Anything sedating, including valerian, kava, melatonin at higher doses, and CBD, for the additive sedation. Kava also carries its own liver signal.

Interactions & What NOT to Take

Dopaminergic drugs

Levodopa (Sinemet), pramipexole, ropinirole, rotigotine, bromocriptine, cabergoline. Direct pharmacological opposition at D1 and D2. Do not combine.

Antipsychotics

Haloperidol, chlorpromazine, risperidone, olanzapine, quetiapine, aripiprazole, ziprasidone, lurasidone. Additive D2 blockade. The concern is extrapyramidal symptoms and additive sedation. Metoclopramide and prochlorperazine, which are D2 antagonists used for nausea, belong on this list.

Sedatives and CNS depressants

Benzodiazepines, zolpidem and eszopiclone, opioids, gabapentin and pregabalin, first-generation antihistamines (diphenhydramine, doxylamine), alcohol. Additive sedation and, at the extreme, respiratory depression, which the pediatric THP cases demonstrated.

CYP3A4 substrates

THP inhibited CYP3A4 in vitro and raised midazolam AUC by 39% to 57% in dogs at 12 mg per kilogram three times daily. A rat study of total Corydalis alkaloids at 6 to 150 mg per kilogram for 14 days induced CYP3A1 and CYP2E1 (Yan and colleagues, BMC Complementary and Alternative Medicine, 2014), the opposite direction. At supplement THP doses neither effect is likely to be measurable in a person, but no human study has looked. Tacrolimus, cyclosporine, sirolimus, and simvastatin are the drugs where even a small shift matters.

CYP2D6 substrates

Low-affinity inhibition of CYP2D6 by THP is reported in vitro; the dog study found no meaningful effect on metoprolol. Low priority.

Drugs that slow the heart

Beta blockers, digoxin, diltiazem, verapamil, ivabradine. The bradycardia in THP overdose was profound; at supplement doses no interaction has been reported, but anyone with a resting pulse under 55 should treat this as a reason not to experiment.

Hepatotoxic drugs and supplements

Methotrexate, isoniazid, valproate, high-dose acetaminophen, kava, green tea extract at high doses, anabolic steroids. The 2025 Chinese pharmacovigilance signal for Corydalis Rhizoma and liver injury is reason enough not to add it to a regimen that already loads the liver.

OATP1B1 substrates

THP inhibited OATP1B1-mediated uptake by more than 50% in transfected HEK293 cells (Sun and colleagues, Chemical Research in Toxicology, 2025), reducing uptake of methotrexate and microcystin. Statins and methotrexate use this transporter. In vitro only.

Warfarin

No data either way. Absence of data is not safety.

Quality, Testing & Adulteration

The 2025 survey, in detail

Luis and Schneider bought 14 products online (four powders or granules, eight capsule products, two liquids), noting that three Nashville retail stores stocked none, and quantified 19 of the 84 reported Corydalis alkaloids by HPLC with diode array detection after weak cation exchange cleanup. Five products (samples 4, 5, 7, 11 and 12) clustered at 9.5 plus or minus 1.6 mg per gram total, near the 12.7 mg per gram they measured in a ground tuber bought online as a reference. Five (samples 1, 3, 6, 13 and 14) sat at 1.8 plus or minus 0.9 mg per gram. Three (samples 8, 9 and 10) were near or below quantification. On repurchase, sample 8 went from nothing to 2.5 mg per gram, sample 9 fell from 0.23 to 0.006 mg per gram, sample 2 lost its 5 mg per gram THP spike, and sample 10 had disappeared from the market. Sample 9's label named DHCB and the product had the least DHCB of any sample. Sample 12, labeled only "Corydalis poppy powder" with no Supplement Facts panel, was among the richest. The authors called the near-empty products fraudulent, while noting that a label reading "Corydalis" without a species may not be legally false. They also stated the method was not formally validated, so treat the numbers as a survey, not a reference assay.

Substitution risk at the botanical level

Corydalis is a genus of several hundred species, and the THP-bearing Stephania species carry an overlapping Chinese name (jin bu huan). A product labeled "Corydalis" could be C. yanhusuo, another Corydalis species such as C. saxicola (used in China for liver disease), or a Stephania. DNA barcoding would settle species; no US product states that it has been done.

Spiking with purified THP

One of 14 products was found with a profile consistent with a spike: a single dominant THP peak with the other alkaloids near absent. A spiked product delivers the sedating, hepatically active alkaloid at pharmaceutical fractions of a dose with none of the matrix. The tell on a label is a THP standardization claim that is suspiciously high for a water extract.

Third-party testing

As of this writing I could not find a USP, NSF or ConsumerLab program that has tested or certified a Corydalis product, and the USP Herbal Medicines Compendium does not appear to carry a Corydalis Rhizoma monograph. The standards that exist are the Chinese Pharmacopoeia (THP not less than 0.050%) and the Hong Kong Chinese Materia Medica Standards (corydaline plus THP not less than 0.10%, HPLC fingerprint with five characteristic peaks, foreign matter not more than 1.0%). Neither sets a DHCB or dehydrocorydaline value, and neither has any force over a US supplement.

What a certificate of analysis needs to say, and how you check it yourself

Ask the seller for the document before you buy. It should name Corydalis yanhusuo, name the tuber as the plant part, give an HPLC value for THP in percent or mg per gram (a legitimate tuber or water extract runs roughly 0.03% to 0.2%; anything over 0.5% is a red flag for spiking, not a sign of potency), ideally give dehydrocorydaline and DHCB values as well, and carry heavy metal and microbial results with the lab named. Without a lab you can still check the species on the Supplement Facts panel, the plant part, whether the ratio claim is internally consistent (a true 20:1 extract should assay far above raw tuber; in the survey it assayed below), and the taste, which for real tuber powder is distinctly bitter.

Special Considerations

The numbers that do not belong in a dosing section, scoped

The rodent doses: synthetic DHCB 5 to 40 mg per kilogram intraperitoneally in male CD1 mice, non-sedating at 10 mg per kilogram and below; water extract 100 to 500 mg per kilogram intraperitoneally in Swiss Webster and 129/sv mice, non-sedating at 200 mg per kilogram. These are injected doses in 20 to 40 g animals and have no oral human equivalent that anyone has established. The Chinese prescription doses: yanhusuo 3 to 10 g per day in decoction, or 1.5 to 3 g per dose as powder, under the 2020 Pharmacopoeia, prescribed by practitioners who also decide what it is combined with; and Rotundine, purified levo-THP, 60 to 180 mg per day, a licensed drug. None of these is a supplement dose.

Regulatory status in the US

Corydalis tuber is sold under DSHEA as a dietary ingredient with no FDA review. Purified THP has no US drug approval and I could find no lawful US supplement use of it. Wang and Mantsch (2012) state that some THP-containing preparations, most notably Jin Bu Huan, are banned by FDA for human consumption, and Luis and Schneider (2025) repeat that "certain preparations" are banned; I could not locate the FDA document or import alert itself, so treat that as a secondary-source statement. What the primary record shows is narrower: the August 1993 MMWR noted that the product's disease claims made it subject to regulation as a drug, and the December 1993 MMWR stated only that FDA was investigating the product.

Older adults

Sedation, bradycardia and polypharmacy converge here. The 66-year-old in the Los Angeles series developed hepatitis on two tablets a night, two or three nights a week.

Surgery

Stop at least a week before any procedure, for the sedation and theoretical CYP3A4 and cardiac interactions with anesthetics.

Research Status & Evidence Quality

Tier 1, replicated animal pharmacology with a defined mechanism

DHCB and the water extract reduce acute thermal, inflammatory and neuropathic pain behaviors in mice, without tolerance over 7 days, through a mechanism that is partly dopamine D2 dependent and not opioid dependent. Two papers from one group (UC Irvine, 2014 and 2016), with the mechanism cross-checked by receptor knockouts and selective agonists. Strong for what it is; it is mouse data with injected drug.

Tier 2, human data on the purified alkaloid

THP has a decades-long prescription history in China at 60 to 180 mg per day, one US placebo-controlled safety study of 60 mg per day for 4 days in 24 men, and two Chinese trials in heroin users (120 and 60 patients) reported by Wang and Mantsch. None of these measured pain.

Tier 3, human data on the herb

One controlled cold pressor trial, 2004, single oral dose, two-herb combination, healthy volunteers, published without the details needed to judge it. Nothing since.

Tier 4, product quality

One peer-reviewed survey of 14 US products (2025) with an unvalidated method, showing that about a third of products matched the raw tuber, a third delivered a fraction, and a fifth delivered nothing or next to nothing.

Research Limitations

No randomized trial of any Corydalis product for any pain condition in humans has been published in English. The analgesia mechanism is established in male mice only. The 2016 study's own finding that purified THP plus DHCB did not reproduce the extract's effect means the active principle of the extract is not fully identified. The main hepatotoxicity record is on a THP tablet product of unconfirmed botanical origin; the handful of case reports on corydalis products (US 2023, China 2021, Korea 2009) did not assay the product or confirm the species, and the Chinese pharmacovigilance signal for the herb itself has not been published with case-level detail in English. There are no pregnancy, pediatric, elderly, renal or hepatic impairment data for the herb. The Chinese Pharmacopoeia marker (THP) is not the compound in the analgesia research (DHCB) and is not the most abundant alkaloid (dehydrocorydaline), so a product "standardized to THP" is standardized to the wrong thing for three different reasons.

Summary & Key Takeaways

Bottom Line

Corydalis yanhusuo has a real pharmacology, a real mechanism and no human efficacy trial of any product you can buy. The compound in the headlines, DHCB, was injected into mice at milligrams per kilogram and is present in a good capsule at under a milligram. The compound the Pharmacopoeia measures, THP, is a sedating prescription dopamine antagonist in China with a poisoning and hepatitis record on purified tablets. The compound actually dominating the powder, dehydrocorydaline, is the one with the least research. Of 14 US products tested in 2025, five were close to raw tuber, five were weak, three were empty or nearly so and one was spiked with THP.

Key Safety Points

• Do not take it in pregnancy, while breastfeeding, with known liver disease, or alongside levodopa or any dopamine agonist.

• Do not combine with antipsychotics, sedatives, opioids or alcohol; expect sedation and test a first dose in the evening.

• Stop at the first sign of liver trouble (fatigue, nausea, dark urine, itching, jaundice) and get enzymes checked; rechallenge caused relapse in the Jin Bu Huan cases and in the 2023 US corydalis sleep-aid case.

• Keep every product away from children; the 1993 pediatric THP overdoses produced respiratory depression and, in two of three, bradycardia within an hour.

• A product labeled only "Corydalis" with no species, or standardized to a suspiciously high THP value, is more likely to be empty or spiked than potent.

Special Note

The Jin Bu Huan cases are the most cited safety fact about Corydalis and are not, strictly, a Corydalis finding: the tablets were a 36% purified alkaloid product whose label named two wrong genera and whose true source was probably Stephania. The reason that history still matters is that the alkaloid was THP, the same compound by which Corydalis is officially measured, that in 2025 a US Corydalis supplement was found carrying 5 mg of it per serving with the rest of the plant missing, and that in 2023 two US adults developed hepatitis on a retail sleep aid listing corydalis as its active ingredient.

Researched and drafted with AI assistance. Every claim verified against primary sources and human-reviewed before publication. Last reviewed: October 15, 2026. Corrections: reply to any issue.

This issue is educational and is not medical advice. If you take any medication, have a diagnosed condition, are pregnant or breastfeeding, or are considering Corydalis for a child, talk to a clinician before changing anything.