Yes, cannabinoids can be detected in human breast milk, and THC is the cannabinoid found most consistently. Research using sensitive laboratory methods has identified THC and its metabolites in milk from mothers who use cannabis, with concentrations ranging from trace amounts to roughly 300 nanograms per milliliter. Detection depends on dose, frequency, route of use, and how recently the mother used. The clinical significance for a nursing infant is still an open question, and no health authority currently endorses cannabis use while breastfeeding.
How Cannabinoids Get Into Breast Milk
Cannabinoids are fat-soluble molecules, and that single chemical property explains almost everything about their presence in milk. Human milk is roughly 3.5 to 4 percent fat, and THC partitions into that fat fraction readily. The same lipophilicity that lets THC cross the blood-brain barrier lets it cross into the mammary gland's milk droplets.
The mechanism runs in three stages. First, inhaled or ingested THC enters the bloodstream and binds plasma proteins, though a meaningful fraction remains free and diffusible. Second, that free fraction crosses the alveolar cell membranes of the breast by passive diffusion, driven by the concentration gradient between plasma and milk. Third, once inside, the drug dissolves into the milk fat globules that form during lactation. In plain terms: THC follows the fat, and breast milk is a fat-rich fluid.
THC is also stored in the mother's body fat and released slowly back into plasma, which is why detectable levels can persist in milk well after the last use. A mother does not need to use cannabis on a given day for milk to contain measurable THC.
What the Numbers Actually Show
Quantified milk concentrations are lower than many people assume, but "low" is not the same as "safe." The table below summarizes concentration ranges reported in the peer-reviewed lactation literature for the two most-studied cannabinoids.
| Cannabinoid | Typical milk range | Most common detection | Key caveat |
|---|---|---|---|
| THC | ~0.03 to 300 ng/mL | Detectable in most users | Rises with dose and recent use |
| CBD | Often undetectable to ~10 ng/mL | Less consistent | Limited data at present |
| THC-COOH (metabolite) | Trace to low ng/mL | Variable | Marker of recent exposure |
For context, a 4-kilogram infant drinking about 750 mL of milk per day at a THC concentration of 30 ng/mL would ingest roughly 22 micrograms daily. At the high end of the range, near 300 ng/mL, ingestion climbs toward 225 micrograms per day. These are micrograms, not milligrams, which is one reason the absolute infant dose is small compared with adult recreational doses measured in single-digit milligrams.
The variable most strongly tied to milk concentration is how recently and how much the mother used. One study of mothers who used cannabis found THC detectable in a majority of milk samples, with concentrations peaking in the hours after use and declining over subsequent days. The half-life in milk is long enough, on the order of days rather than hours, that "pumping and dumping" for a single feed does little to clear it.
A Timeline of Cannabis and Lactation Research
Understanding what is known requires knowing when it became known. The evidence base is younger than most people realize.
| Period | Milestone |
|---|---|
| 1980s | Early case reports note cannabinoids in breast milk, but methods are insensitive |
| 1990 | Landmark case report (Perez-Reyes and Wall) confirms THC transfer at low levels |
| 2010s | Sensitive mass spectrometry enables reliable detection and quantification |
| 2018 | Larger lactation studies show wide concentration variability between mothers |
| 2020s | Reviews confirm frequent detection, but infant outcome data remain sparse |
| 2022 | Professional bodies publish cautious guidance against use while nursing |
| 2026 | Still no established safe threshold for infant exposure |
The pattern is clear: detection technology advanced faster than outcome science. Researchers can now measure cannabinoids in milk with great precision, but long-term infant neurodevelopmental data are still limited and largely observational.
Infant Exposure and the Blood-Brain Barrier Question
The infant's own physiology changes the calculation. A newborn's liver enzymes, particularly the CYP2C and CYP3A families responsible for metabolizing THC, are immature for the first months of life. Oral bioavailability is also higher in infants than adults because first-pass metabolism is less efficient. A dose that would be trivial for an adult stomach may reach an infant's bloodstream in a larger fraction.
The counterintuitive part: the infant's greater vulnerability comes less from the size of the dose than from the immaturity of the clearance system. A small amount of THC that an adult clears in hours may linger far longer in a neonate. This is a hypothesis supported by pharmacokinetic principles and by what is known about other lipophilic drugs, but direct infant THC pharmacokinetic studies are essentially absent.
One professional caveat deserves emphasis. Detection of a cannabinoid in milk proves exposure, not harm. The two questions, whether the drug transfers and whether the transfer causes measurable injury, are separate, and the second remains understudied. Statements that cannabis "is safe" in lactation are not supported by the evidence, and neither are claims of proven severe harm.
Misconceptions and Edge Cases
Several recurring beliefs about cannabis and breastfeeding do not survive scrutiny.
Pumping and dumping does not work for THC the way it does for alcohol. Alcohol clears from milk as blood alcohol falls, tracking a predictable curve. THC is stored in fat and released over days, so discarding milk for a few hours removes very little of the total exposure.
CBD products are not automatically safer. Many CBD oils contain variable THC, and labels are often inaccurate. A mother using a "THC-free" CBD tincture may still expose her infant to unlisted cannabinoids.
Route of use matters but does not eliminate risk. Vaping and smoking deliver THC rapidly and produce peaks, while edibles produce lower, more sustained plasma levels. Lower peaks do not mean zero transfer, because the fat depot keeps releasing drug.
Secondhand smoke exposure is its own pathway. An infant in a room where cannabis is smoked inhales cannabinoids directly, independent of milk. This is a separate exposure route that milk-focused discussions often omit.
A genuine edge case is the mother using prescribed cannabinoid medication for a serious condition such as severe nausea. Here the calculation shifts, because the alternative may be uncontrolled illness. These situations require individualized clinical input rather than a blanket rule.
What the Guidance Says and What to Actually Do
No major health body endorses cannabis use during breastfeeding. The American College of Obstetricians and Gynecologists advises against it, and the American Academy of Pediatrics has recommended that mothers avoid cannabis while nursing. Per the CDC, the safest course is to avoid cannabis during lactation until better data exist. These positions reflect precaution in the absence of proven safety, not proof of a specific harm threshold.
For a mother who wants concrete steps rather than a warning label, the following actions are reasonable.
- Discuss any cannabis use with the infant's pediatrician and the mother's own clinician, disclosing dose and frequency.
- Avoid smoked or vaped cannabis around the infant to prevent secondhand exposure, regardless of feeding method.
- Do not rely on pumping and dumping as a protective strategy, because it does not meaningfully reduce stored-drug exposure.
- Choose regulated products with accurate labels if use is unavoidable, since unverified CBD and THC content is a recurring problem.
- Consider the infant's age, because preterm and newborn infants clear drugs less efficiently than older babies.
- Reassess use as circumstances change, since occasional use carries a different risk profile than daily use.
Limits of the Evidence and Open Questions
The honest summary is that transfer is established and outcome data are thin. Milk concentrations vary by more than three orders of magnitude across studies, from under 1 ng/mL to roughly 300 ng/mL, and no study has defined a concentration below which infant effects can be ruled out. Reconciling the thresholds: detection is common above about 1 ng/mL, meaningful infant intake emerges at the higher end of that range, and a universally safe level remains unestablished.
Three questions would settle much of the uncertainty. First, what are the actual plasma levels achieved in breastfed infants? Second, does chronic low-level exposure affect neurodevelopment over years? Third, do route and product type change infant outcomes, or only maternal milk levels? Until these are answered with longitudinal data, the science supports caution, precise language about what is known, and individualized clinical advice rather than false reassurance in either direction.