Is BPA-Free Safe? The Evidence on BPS, BPF, and Replacements

Usually it isn't. The chemical that replaced BPA in most of those bottles binds the same estrogen receptors BPA does, at roughly the same strength, through a signalling pathway that regulatory safety testing wasn't looking at until 2013. The substitutes — BPS, BPF, BPAF — behave like BPA because they're shaped like BPA. The 'BPA-free' label was a regulatory exemption, not a safety improvement.

The substitutes have names you've probably never seen on a label. BPSBisphenol S — a chemical analogue of BPA used as the most common substitute in thermal receipts and many BPA-free plastics. Structurally similar enough to bind the same estrogen receptors. (bisphenol S), BPF (bisphenol F), BPAF (bisphenol AF) and several close cousins now share the BPA-shaped role in plastics, can linings, dental sealants and thermal paper. Our profile of BPA itself covers the molecule's backstory — this article is about what came after it.

The question of whether any of the replacements were actually safer got its answer in 2011. Yang and colleagues took commercial plastic products straight off the shelf — baby bottles, food containers, deli packaging — extracted the chemicals that leached out of them, and dosed the extracts onto MCF-7 human breast cancer cells, which proliferate in response to anything that acts like estrogen. Almost every product tested made the cells grow. Including the ones advertised as BPA-free Yang et al. 2011. The follow-up three years later extended the pattern to specifically BPA-free replacements Bittner et al. 2014 — more on that paper in a moment. Disclosure — Yang and Bittner are associated with CertiChem and PlastiPure, commercial EAestrogenic activity testing labs. The findings have been argued about, but the pattern has been reproduced by independent researchers.

When regulators restricted BPA, industry needed structurally similar molecules that could do the same chemical jobs — monomers in polycarbonate-type plastics, cross-linkers in epoxy resins, colour developers in thermal paper. The bisphenol family is large. Manufacturers picked the closest cousins. Most of them retain the same double-phenol-ring structure that binds estrogen receptors, because that structure is exactly what lets BPA do its industrial work in the first place.

BPS has the cleanest cautionary tale. Researchers raised alarms about BPA coming off thermal receipts — the slippery ones you get at the supermarket till — and regulators moved. France banned BPA from food contact materials in Loi n° 2012-1442, in force from 2015. The EU followed with a dedicated thermal-paper restriction (Commission Regulation 2016/2235) in 2020. Manufacturers complied by swapping BPA for other colour developers, and the obvious chemical choice was BPS.

Liao and colleagues had already seen what that would look like. In 2012 they collected 111 thermal receipts from shops and ATMs across four countries thermal receipts from shops and ATMs across the US, Japan, Korea and Vietnam and tested every one of them for BPS.

They found it in all of them. Every receipt. Every country. Geometric-mean concentrations of 0.181 mg/g — the same order of magnitude as the BPA those same receipts had been printed with a few years earlier Liao, Liu, Kannan 2012. Eckardt and Simat followed the German market as the ban took hold from 2015 to 2017 and watched the picture fragment: BPA still dominant at the start, BPS climbing, Pergafast 201 catching up to about a third of samples by the end of the window Eckardt and Simat 2017. The chemical changed. The exposure didn't.

The most-cited systematic review of BPS and BPF pooled 32 in vitro and in vivo studies and reached a blunt conclusion: both substitutes show estrogenic, anti-estrogenic, androgenic and anti-androgenic activity in the same order of magnitude as BPA, across multiple receptor pathways and assay types Rochester and Bolden 2015. The potency profiles weren't identical — BPF skewed slightly more estrogenic in some assays, BPS slightly less at the nuclear receptor — but the summary holds. Both chemicals act like BPA because they're shaped like BPA.

The paper that matters most for understanding BPS at the cellular level came the year before the review. Viñas and Watson took GH3/B6/F10a rat pituitary tumour cell line widely used to study membrane-initiated estrogen signalling rat pituitary tumour cells — the kind routinely used to watch estrogen signalling happen in real time — and dosed them with BPS at vanishingly small concentrations. Femtomolar. Nanomolar. Levels well below anything a regulatory safety test would think to check for. The cells responded the way they respond to estradiol, the body's own primary estrogen Viñas and Watson 2013. Not through the nuclear estrogen receptor — which is the one industry and regulators point at when they call BPS weaker than BPA — but through the membrane-bound pathway running alongside it. A receptor that wasn't on the safety-testing checklist.

The 'weaker' story was true of the receptor they'd looked at. It missed the receptor they hadn't.

The pattern extends beyond BPS. A 2024 in-vitro screen of eleven common BPA analogues — run by Reininger and Oehlmann and titled, with the question mark intact, 'Regrettable substitution?' — found effect profiles broadly comparable to BPA itself, and in some cases worse Reininger and Oehlmann 2024. None of this is new. Kitamura's group had already compared BPA against nineteen structurally related compounds back in 2005, flagging several — notably TCBPA and bisphenol B — as more estrogenic than BPA, plus thyroid-hormone activity from TBBPA, TCBPA and TMBPA Kitamura et al. 2005.

Five years before 'BPA-free' went mass-market.

BPAF is a slightly different case. It isn't just BPA with a few extra atoms — it acts as a full agonist for estrogen receptor alpha and a highly specific antagonist for estrogen receptor beta, binding ERα around twenty times more effectively than BPA and ERβ about fifty times more effectively Matsushima et al. 2010. The two receptors normally push in opposite directions on cell proliferation, so a chemical that switches one on while blocking the other is pharmacologically distinctive in a way BPA is not. When the US National Toxicology Program ran a modified one-generation reproductive study on BPAF in rats — published as NTP DART-08 in 2022 — the findings came back as 'clear evidence of reproductive toxicity and clear evidence of developmental toxicity' at dietary exposures: F1 reproductive failure, fetal malformations, reproductive-tract histopathology DART-08 is a developmental and reproductive study, not a two-year cancer bioassay — the popular framing of NTP's BPAF work as a carcinogenicity finding has circulated widely but doesn't match the published report..

The pattern — ban one chemical, replace it with a structurally similar untested cousin — has a name in the chemical policy literature: regrettable substitution. It happens because the regulatory framework evaluates chemicals individually rather than by structural class. A restriction on BPA doesn't automatically trigger review of the other bisphenols. A restriction on PFOA doesn't require testing of the thousands of other PFAS compounds in the family. The manufacturer's legal obligation ends at removing the named chemical, and the clock starts fresh on whatever replaces it.

Bittner, Yang and Stoner made the argument directly in the title of their 2014 paper: estrogenic chemicals often leach from BPA-free plastic products. Over a thousand extracts from around fifty BPA-free replacement items went through the same MCF-7 assay as the 2011 work, this time stressed the way a container actually gets stressed — heat from a dishwasher, a few hours in a sunlit car, UV exposure, microwaving. Many of them, especially polyethersulfone, polystyrene and certain Tritan-style resins, still released chemicals with estrogenic activity after the stress. Several PETGpolyethylene terephthalate glycol — a thermoplastic resin sometimes used as a BPA-free replacement that tested EA-negative in Bittner 2014 and cyclic-olefin products came out clean. Worth knowing. The broader point wasn't that every BPA-free replacement is worse than BPA. It was that the label on the bottle can't tell you which is which.

The 'BPA-free' claim, in this framing, is a label-removal exercise. The chemical you've heard of is gone. The chemical you haven't heard of has taken its place. The statement is technically accurate — the product genuinely contains no BPA — and substantively misleading for most of the replacements, because the substitute does roughly the same biological thing at the same receptors. The system worked exactly as designed. The design wasn't built to catch substitution within a structural class.

Sort of — and only for food contact materials, and only in one jurisdiction. The European Commission adopted Regulation (EU) 2024/3190The 2024 EU regulation amending Commission Regulation (EU) No 10/2011 on plastic food contact materials, banning BPA and other hazardous bisphenols from food contact. on 19 December 2024, and it entered force on 20 January 2025. The regulation amends the core EU plastic food contact rulebook (10/2011) and replaces the earlier 2018 BPA-specific migration limit. What it bans is BPA itself, plus any 'hazardous bisphenols' that meet carcinogenic, mutagenic, reprotoxic or endocrine-disrupting criteria under the CLPClassification, Labelling and Packaging framework — which in practice captures BPS, BPAF, TBBPA, phenolphthalein and 4,4'-isobutylethylidenediphenol. It is the first serious jurisdictional move to treat the bisphenol class as a class rather than chasing one molecule at a time.

The phase-in is staggered over several years. General single-use and repeat-use food contact articles have a first deadline of 20 July 2026. Repeat-use professional food production equipment gets 20 January 2028. Single-use articles with varnishes or coatings for fruit, vegetables and processed fish get 20 July 2028. Repeat-use articles already on the market have until 20 January 2029, with a further twelve months during which food inside them can still be sold. The regulation doesn't reach thermal receipts — those are covered separately by Commission Regulation 2016/2235, which restricted BPA (but not its replacements) from 2020.

Outside the EU food-contact scope, nothing much. The UK Food Standards Agency opened a 12-week consultation in October 2025 on aligning Great Britain with the EU position; Northern Ireland is already bound through the Windsor Framework, but as of 2026 Great Britain has not laid the legislation. The US Food and Drug Administration has not introduced anything comparable. The FDA's BPA restrictions remain limited to baby bottles, sippy cups and infant formula packaging, and in October 2024 the agency reaffirmed its position that current BPA exposure levels in food contact materials are safe. No federal rulemaking on BPS, BPF or BPAF is pending.

The honest answer is that 'BPA-free' on a label tells you almost nothing on its own. What works is avoiding the categories of product where bisphenols live, regardless of what the label says. Glass, stainless steel, ceramic and silicone for food storage — not 'BPA-free plastic,' which usually means BPS or BPF plastic. Fresh or frozen food rather than canned, where possible. Thermal receipts handled with dry hands, or politely declined. Product-level certifications rather than marketing copy on the front of the bottle.

If you're in the EU and shopping for food packaging from 2026 onwards, the legal situation is finally improving — Regulation 2024/3190 is phasing in, and the food contact ban will eventually catch most of the substitutes. If you're in the UK or the US, you're still on your own. Either way, the safest working assumption is that any plastic in contact with food, drink or your skin contains some kind of bisphenol unless the product specifically and verifiably says otherwise — and that the bisphenol it contains behaves roughly like the BPA you were trying to avoid in the first place.

The green badge on the front of that water bottle is the textbook case of how regulatory chemistry can produce a technically accurate statement that misleads everyone reading it. The chemical the consumer recognised was removed. The chemical that took its place shared the same biological target, the same broad effect profile, and a smaller body of safety research — because nobody was incentivised to produce the research that might kill its market.

The evidence has been sitting there for more than a decade. Rochester and Bolden published their systematic review in 2015. The 2024 EU regulation finally caught up to it, for one category of product, in one jurisdiction. Outside of that, the best consumer move is the same as it was before the badges went on the bottles: avoid the categories where bisphenols live, regardless of what the label says. Glass and steel never had a label-removal exercise to do, because they never contained BPA in the first place.