In April 2023, two of the most capable food safety agencies in the world published their conclusions about the same chemical. One said it is safe at the amounts people get from food. The other cut its tolerable daily intake by a factor of roughly 20,000 and said typical European diets now exceed it.
The European Food Safety Authority, EFSA, published its re-evaluation on 19 April 2023. The US Food and Drug Administration’s consumer page on bisphenol A carries a content-current date of 20 April 2023, the very next day. That FDA page names the National Toxicology Program, the EPA, the Consumer Product Safety Commission, the CDC, Health Canada and the World Health Organization. It does not mention EFSA.
Nobody here is lying. Two sets of competent scientists read the same pile of studies and made different judgment calls at four or five specific points, and those calls compounded.
This article will not tell you which agency is right. It tells you what each one decided, when, what the decision rests on, and what would have to be true for each to hold up.
Where BPA shows up in your day
Bisphenol A is a building block, not a product. According to the FDA, it is used mainly to make polycarbonate plastic, the hard clear kind that resists shattering, and epoxy resins, which include the thin coating sprayed inside most metal food and drink cans. The National Toxicology Program describes the same two uses.
Those two jobs explain most of what reaches you. Canned food sits against an epoxy lining for months. Polycarbonate turns up in reusable water bottles, some kitchen gear and the clear housings on appliances.
Then there is a route most people never think about. BPA is the most common color developer in thermal paper, the shiny stock used for register receipts, parking tickets and shipping labels. It is not bound into the paper. It sits on the surface and transfers to skin.
How much do Americans take in? A 2015 analysis in Environmental Research, using the 2011 to 2012 National Health and Nutrition Examination Survey, estimated a median daily intake for the US population of about 25 nanograms per kilogram of body weight, and found intakes had fallen significantly since the 2003 to 2004 round. The authors also spent much of the paper on why such numbers are hard to pin down: a spot urine sample captures only the last several hours, and the methods used to correct for dilution give different answers.
Hold on to that figure. It is about to do a lot of work.
The FDA’s one-sentence position, and the study behind it
The agency’s position is one sentence, and it has not softened it: the FDA states that “BPA is safe at the current levels occurring in foods.” The page adds that the available information continues to support the safety of BPA for its currently approved food-contact uses.
The reasoning behind that sentence is mostly about what your body does with the chemical, not about what the chemical does to cells in a dish. The FDA says oral BPA is converted quickly into an inactive form, and its physiologically based models predict that internal exposure to the active form after swallowing is 1 percent or less of the dose taken in. Primates, including people, clear it faster and more efficiently than rodents do. In one study at the FDA’s National Center for Toxicological Research, pregnant rodents were dosed orally with 100 to 1,000 times more BPA than people get from food, and the active form could not be detected in the fetus eight hours later.
The biggest piece of that work has a name. CLARITY-BPA, short for the Consortium Linking Academic and Regulatory Insights on Bisphenol A Toxicity, was launched jointly by the National Institute of Environmental Health Sciences, the National Toxicology Program and the FDA. Its core study was a two-year, guideline-compliant rat study run at the FDA’s center, with doses spanning 2.5 to 25,000 micrograms per kilogram of body weight per day, a four-order-of-magnitude range. University researchers ran additional studies on blinded samples, so they did not know which animals got which dose.
The core results appeared in Food and Chemical Toxicology in 2019. The authors reported no BPA-related effects in the clinical and non-pathology data. The tumors and tissue changes they found were the ordinary lesions of aging rats, scattered across control and treated groups without a consistent pattern. They flagged one possible relationship, in the female reproductive tract and the male pituitary, and only at the highest dose of all: 25,000 micrograms per kilogram per day.
That is the evidence tier the FDA is working from: a large, deliberately conventional study built to the rules regulators normally use, which did not produce a consistent signal except at the top of a very wide dose range.
What Europe’s food regulator decided in April 2023
EFSA’s numbers tell a story by themselves. Its tolerable daily intake for BPA was 50 micrograms per kilogram of body weight per day. In 2015 it dropped that to a temporary intake of 4 micrograms per kilogram per day. In the 2023 re-evaluation, EFSA states that it established a new TDI of 0.2 nanograms per kilogram of body weight per day.
Read the units twice. Micrograms became nanograms. EFSA’s plain-language summary spells out that 0.2 nanograms is two tenths of one billionth of a gram, and that the new figure is about 20,000 times lower than the 2015 value.
The basis is an immune effect. EFSA’s announcement describes an increase in a type of white blood cell, T helper cells, in the spleen, a change the panel judged could lead to allergic lung inflammation and autoimmune disease. The panel also looked at reproductive, developmental, metabolic and neurobehavioural effects. On top of its usual safety margins, it applied an additional uncertainty factor of 2 to cover uncertainty in pinning down the toxicological starting point.
Then the arithmetic. EFSA concluded that dietary exposure runs two to three orders of magnitude above the new TDI, and that consumers with both average and high exposure, in every age group, exceed it. Its exposure estimates come from the 2015 assessment and do not account for restrictions imposed since.
Set that against the US figure. A median American intake of about 25 nanograms per kilogram per day is roughly 125 times EFSA’s 2023 TDI. The same measurement sat two orders of magnitude below the benchmark in force when it was published and now sits more than two orders of magnitude above the one EFSA uses. The number did not change. The yardstick did.
The European Commission acted on the opinion. On 19 December 2024 it adopted a ban on BPA in materials that touch food or drink, including can coatings and reusable bottles, with an 18-month phase-out for most products.
Why two competent agencies landed so far apart
You do not have to guess at the substance of the disagreement. EFSA published a joint report with Germany’s Federal Institute for Risk Assessment, known as BfR, which took part in the review and did not agree with its conclusions. That document is unusually candid, and it puts the argument in three places.
Which changes count as harm
EFSA treated the rise in splenic T helper 17 cells as adverse. BfR said there was no convincing evidence tying that marker to an actual adverse outcome, judged it unsuitable for setting a health-based guidance value, and called its use a shift in approach.
This is the heart of it, and it is not really a dispute about data. A cell-population shift is an intermediate endpoint: a biological change somewhere on a pathway, not a disease. Whether a pathway change counts as harm is a judgment call about biology, not a measurement. EFSA’s position is that intermediate endpoints can be adverse when a clear causal link to disease exists, and it pointed to work connecting T helper 17 pathways to inflammatory disease. BfR’s position is that the role of these cells is context-dependent and incompletely understood, that the animal disease-model studies are of low quality, and that human epidemiology is missing.
One detail explains a great deal. The joint report notes that CLARITY-BPA did not specifically examine T helper 17 cells in mice. The big guideline study the FDA leans on was not built to test the endpoint EFSA built its number on. The two bodies were not weighing the same measurement differently. They were weighting different measurements.
How much uncertainty to build in
EFSA ran a formal expert elicitation and justified an extra factor of 2. BfR argued that stacking worst-case assumptions at every step produces an over-conservative value, that the uncertainty analysis should rest on observed data rather than expert judgment, and that the resulting TDI came out several orders of magnitude lower than it expected. EFSA’s reply was that every scientific assessment involves expert judgment.
They also split on converting animal dose to human dose. BfR argued the realistic factor is 10 to 100 times higher than the one EFSA used, and that the study EFSA based it on was unsuitable. EFSA rejected the alternatives but folded a possible tenfold higher factor into its revised uncertainty analysis.
Which studies belonged in the pile
BfR judged one ovarian study unreliable and would have dropped it; EFSA kept it at a lower confidence tier, and both agreed from the overall weight of evidence that ovary effects existed. BfR also objected to the window of publication dates EFSA searched and would have downgraded studies over chemical purity and background contamination. EFSA answered that it published its protocol in advance, in 2017, and judged contamination negligible.
Where did this leave them? The joint report is blunt. BfR and EFSA concluded that “it is not possible to achieve convergence” on the hazard characterisation for BPA, and agreed only to keep talking about how to align their methods.
That sentence should change how you read the gap. It does not close by reading more carefully. It turns on prior commitments about what counts as evidence of harm, and those were in place before either body opened a single study.
The baby bottle story runs backwards in most people’s heads
Ask people why BPA disappeared from baby bottles and most will say the FDA banned it because it was dangerous. The record says something else, and you can read it yourself.
Section 177.1580 of Title 21 of the Code of Federal Regulations authorizes polycarbonate resins for food contact, then carves out an exception: the resins may be used under that section except in infant feeding bottles, and except in the spill-proof cups, lids and closures sold to help babies and toddlers learn to drink. The amendment that added the carve-out appeared in the Federal Register on 17 July 2012.
Who asked for it? The American Chemistry Council, the plastics industry’s own trade association. Manufacturers had already stopped. US food additive regulations can be amended to drop a use that has been permanently and completely abandoned, and the FDA says plainly that these amendments are not based on safety, because safety is not what an abandonment determination decides.
A year later, in July 2013, the same mechanism removed the authorization for BPA-based epoxy coatings in infant formula packaging. That petition came from a different direction: Representative Edward Markey of Massachusetts.
The distinction is worth keeping. An abandonment removal tells you what the market did. A safety ban tells you what a regulator concluded. Read the first as the second and the timeline comes out wrong, because it makes the FDA look like it changed its mind in 2012, which it did not.
What BPA-free tells you, and what it does not
Start with what kind of claim it is. BPA-free is not defined in any regulation, it is not a government certification program, and no third party audits it. It is a marketing statement about one named molecule, and the only thing it promises is the absence of that molecule.
What took its place is usually a close relative. Bisphenol S and bisphenol F are the common substitutes, and they are structural cousins of BPA rather than a different kind of chemistry.
The honest state of the evidence is thinner than either side of the argument admits. A 2020 review by researchers at the Netherlands National Institute for Public Health and the Environment catalogued 99 candidate BPA alternatives and narrowed to 20 on the basis of real consumer use. For several of those, data on carcinogenicity, reproductive toxicity and endocrine activity was very limited or absent outright. A 2024 critical review in Critical Reviews in Toxicology, written for a European research partnership, reached the same place: the toxicology of the alternatives now entering products is largely unknown.
Be careful what that means. Thinner data is not evidence of greater harm. It is also not evidence of safety. It is the statement that nobody has looked as hard, which is a different and more awkward thing.
Where data do exist, they do not all run one way. A 2021 study in Annals of Work Exposures and Health measured absorption through human skin samples over 24 hours and found about 25 percent of an applied BPA dose crossed, against roughly 0.4 percent for BPS. For the receipt in your hand, that substitution looks like an improvement. Nobody has run the equivalent comparison for swallowed doses across the whole family of replacements, which is exactly the gap those reviews describe.
Article 718 covers endocrine-disruption claims in general. For this family, the short version is that the label tells you about one ingredient and nothing about the product.
Lowering your exposure, if you decide you want to
Nothing in either agency’s conclusion makes this urgent for most adults. If you want to reduce exposure anyway, these are the levers that move it, ordered by how much they do for how little effort.
- Change what you heat food in, not what you store it in. Heat is the condition that drives chemicals out of a container and into food. Cold storage in the same container is a much smaller exposure. Article 712 covers food-contact materials and microwaving in detail.
- Shift part of your diet off can linings. Fresh and frozen produce does this without a special purchase or a premium label.
- Decline the receipt or take the digital version. Free, takes a second, and removes a skin route entirely. It matters most for people who handle receipts for a living.
- Replace hard plastic containers when they look worn, cloudy, crazed or deeply scratched, rather than on a schedule somebody sold you.
- Skip the chemical panels. No clinical BPA test tells you anything you can act on, because a urine measurement reflects roughly the last several hours rather than your long-run exposure.
If you are pregnant or feeding an infant and this question is genuinely bothering you, the people to ask are your obstetrician, midwife or pediatrician, and the useful question is about feeding equipment and formula packaging rather than about a blood test. See a clinician promptly if an infant has a feeding problem, poor weight gain or a persistent rash you are trying to explain. If a child swallows a household product while you are reorganizing cupboards, call Poison Help at 1-800-222-1222, staffed 24 hours a day. Call 911 for breathing trouble or loss of consciousness.
What would have to be true for each regulator to be right
This is the part worth keeping.
For the FDA’s conclusion to hold, swallowed BPA has to be cleared fast enough that internal exposure to the active form really does stay near 1 percent of the dose. A long, guideline-compliant study has to be a fair test of the effects that matter, including immune ones. And the effects reported only at low doses in non-standard designs have to be either artifacts or changes that are real but not harmful.
For EFSA’s conclusion to hold, a rise in splenic T helper 17 cells has to be a genuine step toward disease in humans rather than a reversible immune fluctuation. The animal-to-human conversion has to be roughly as EFSA set it, not the 10 to 100 times looser figure BfR argued for. And the non-standard studies EFSA retained have to be reliable enough to carry a number this consequential.
Every one of those conditions is testable. None has been tested in a way that settles it. What would settle it is reasonably clear: human studies with an actual immune outcome rather than a cell count, an agreed conversion factor between rodent and human internal doses, and replication of the low-dose immune findings inside the kind of study design regulators already accept.
Until then, the practical position is not fence-sitting. It is this: your exposure is measurable, small in absolute terms, falling in the United States, above one of the two benchmarks in play and well below the other. All of those are true at once. The fight is over where to draw a line, not over the measurement, and a reader who sees that is reading this subject better than most of the people writing about it.
Sources
- U.S. Food and Drug Administration. “Bisphenol A (BPA): Use in Food Contact Application.” fda.gov
- European Food Safety Authority. “Plain language summary: Re-evaluation of the risks to public health related to the presence of bisphenol A (BPA) in foodstuffs.” efsa.europa.eu
- European Food Safety Authority. “Bisphenol A in food is a health risk (19 April 2023).” efsa.europa.eu
- European Food Safety Authority. “Bisphenol (topic page, including the TDI timeline from 2006 onward).” efsa.europa.eu
- European Food Safety Authority and German Federal Institute for Risk Assessment (BfR). “Joint EFSA-BfR report on the re-evaluation of bisphenol A, April 2023.” efsa.europa.eu
- Electronic Code of Federal Regulations. “21 CFR 177.1580, Polycarbonate resins (including the infant feeding bottle and spill-proof cup exclusion).” ecfr.gov
- National Toxicology Program. “Bisphenol A (BPA) and the CLARITY-BPA research program.” ntp.niehs.nih.gov
- European Commission. “Commission adopts ban on bisphenol A in food contact materials (19 December 2024).” food.ec.europa.eu
- PubMed. “Camacho L, et al. "A two-year toxicology study of bisphenol A (BPA) in Sprague-Dawley rats: CLARITY-BPA core study results." Food and Chemical Toxicology, 2019. DOI: 10.1016/j.fct.2019.110728.” doi.org
- PubMed. “LaKind JS, Naiman DQ. "Temporal trends in bisphenol A exposure in the United States from 2003-2012 and factors associated with BPA exposure." Environmental Research, 2015. DOI: 10.1016/j.envres.2015.06.013.” doi.org
- PubMed. “Reale E, Vernez D, Hopf NB. "Skin Absorption of Bisphenol A and Its Alternatives in Thermal Paper." Annals of Work Exposures and Health, 2021. DOI: 10.1093/annweh/wxaa095.” doi.org
- PubMed. “den Braver-Sewradj SP, van Spronsen R, Hessel EVS. "Substitution of bisphenol A: a review of the carcinogenicity, reproductive toxicity, and endocrine disruption potential of alternative substances." Critical Reviews in Toxicology, 2020. DOI: 10.1080/10408444.2019.1701986.” doi.org
- PubMed. “Mhaouty-Kodja S, et al. "A critical review to identify data gaps and improve risk assessment of bisphenol A alternatives for human health." Critical Reviews in Toxicology, 2024. DOI: 10.1080/10408444.2024.2388712.” doi.org