A Fifth of Brazil's Approved Alternative Tests Still Use Animals

Brazil’s register of alternative test methods is 46 entries long. The number is quoted by the regulator, by the industry bodies, by the laboratories that sell against it, and by almost every supplier presentation that mentions the country. Nine of the entries are animal tests.

They are specific and they are not obscure. OECD test guidelines 420, 423, 425 and 129 for acute oral toxicity. Guidelines 429, 442A and 442B for skin sensitisation by local lymph node assay. Guidelines 421 and 422 for reproductive and developmental screening. Nine of 46 is 19.6 per cent — just under a fifth of the register — and none of those nine replaces an animal. They refine it or they reduce the number used. That is a legitimate 3Rs purpose. It is an entirely different commercial object from a replacement method, and any supplier who markets the register as 46 non-animal methods is overselling by a fifth.

This matters more than a counting quibble because the register is not a scientific scoreboard. It is the competitive boundary of a market. Only an assay that CONCEA has recognised can be used to satisfy a Brazilian regulator, and only a laboratory holding the accreditation for that specific method can sell it. The register is the map of what is sellable, so its composition is the market.

The register is deep at the surface and empty inside the body

Read it by endpoint rather than by date and the shape becomes obvious. Brazil’s in vitro capability covers the outside of the body with unusual completeness. It covers almost nothing that has to be established inside the body over time.

Three of those lines are the whole argument. Repeated-dose systemic toxicity has no recognised alternative method in Brazil — none at all. Carcinogenicity has none. Respiratory sensitisation has none. The dermal and ocular battery is finished as a scientific question and is now a cost competition over imported tissue and technician time. The systemic battery has not started.

Which means three ingredient classes cannot be served

Preservatives, UV filters and hair dyes are the ingredient classes where a cosmetic dossier most often has to establish a systemic effect rather than a surface one. They are also the classes the register cannot reach. Everything the market can sell sits on one side of that line, and every ingredient that has to cross it will be tested on an animal for as long as the line holds.

That is where the friction between now and 2031 will sit. Section 16 grandfathers products and ingredients tested before the law took effect, which covers almost everything already on a Brazilian shelf. Section 18 permits a derogation in exceptional cases. What is left is the uncomfortable middle: a genuinely new preservative, with no validated non-animal route and no waiver that fits comfortably.

And the derogation clause deserves to be quoted rather than paraphrased, because it is routinely described as a loophole when it is the opposite. It applies only where the ingredient is widely used and cannot be functionally replaced, a specific human-health problem has been detected, and there is no alternative method capable of meeting the testing requirement — in the statute’s words, “inexistir método alternativo hábil a satisfazer as exigências de testagem.”

That third condition is not a drafting oversight by an industry-friendly legislature. It is an accurate description of an unsolved scientific problem, written into law because the people writing it knew the problem existed. The route out is not a better waiver. It is a test guideline for a microphysiological system, which is why the organ-on-chip pipeline matters commercially here rather than rhetorically — and why it does not matter yet. Until a guideline exists, a chip is evidence, and evidence is not what a dossier needs.

Three compliance clocks have already run out

The register accumulates through normative resolutions, and each one starts its own five-year clock. Normative Resolution 54/2022 gives regulated institutions up to five years from the publication of each recognition resolution to comply with the methods it recognises. Compliance is therefore not recognition, and the register is best read as a queue of future obligations rather than a statement of present capability.

Three of those obligations are already enforceable and nobody in the market behaves as though they are. That is the first thing a new entrant gets wrong in Brazil: the binding requirement is usually the oldest recognition, not the newest announcement.

The next one lands in October 2027, roughly ninety days after the health authorities’ own statutory deadline of 31 July 2027. Two clocks, ninety days apart, and they govern different things — one creates the state’s enforcement machinery, the other makes 16 recognised methods obligatory for regulated institutions. After that, nothing until January, February and July 2031.

Which is why the growth curve is a step, a plateau and a step

A calendar like that cannot produce a smooth curve, so the forecast does not pretend to one. Pheonix Research sizes Brazil’s in vitro toxicology testing market at USD 144.7 million in 2025, reaching USD 286.8 million by 2033 at 8.9 per cent. The shape underneath the average is what a buyer needs: 14 per cent in 2027 as the deadline and the 2022 batch coincide, 5, 4 and 6 per cent through the plateau years of 2028 to 2030, then 16 per cent in 2031 when the 2026 recognitions bind.

That plateau is not a modelling artefact. It is the direct consequence of a three-year gap between two deadlines, and it creates a genuinely unattractive investment problem: a laboratory that wants to serve the 2031 obligations has to build from roughly 2028 and carry three years of cost with no compliance-driven revenue attached. The firms that build early will earn a scarcity premium in 2031. The ones that wait will buy equipment into a queue.

The ban sells nothing. The label sells everything.

Here is the part of this market that is most often misread. The prohibition is not what makes a Brazilian brand buy a reconstructed-epidermis assay. The label is.

Section 15 bars a manufacturer whose safety case rests on new animal data from describing the product as “não testado em animais”, “livre de crueldade” or anything similar. In a beauty market worth R$175 billion in 2025 — the third-largest consumer market in the world, exporting USD 1.061 billion to 189 countries — that claim is not decoration. It is a marketing asset with a measurable value, and section 15 is the only clause in the law that creates demand for testing the ban does not already require.

Section 13, meanwhile, sets the ceiling. Animal data generated after entry into force cannot authorise a cosmetic, except where it was obtained to satisfy a non-cosmetic rule, national or foreign. That exception is not settled by the text, and each generous reading of it removes a dossier from the addressable pool. A preservative tested for a household-chemical registration and then used in a cream is inside it. An ingredient tested for a foreign agrochemical authority and sold into a cream is arguably inside it.

So the clause that governs the top of the forecast is not the prohibition. It is the exception beside it, and how hard ANVISA polices it. Article 2(II) commits the authorities to publishing biennial reports stating how many times documentary evidence of non-cosmetic purpose was requested and how many times companies relied on it — which is an unusually honest legislative instrument, because it obliges the state to publish the number that reveals whether it is enforcing.

How the number was built

One published volume series, one stated price. The volume is ANVISA’s own count of product decisions: 87,706 dossiers regularised in 2022, comprising 73,999 cosmetics, 4,435 saneantes, 7,811 new medical devices and 1,461 agrochemical toxicological evaluations. The price is the average in vitro toxicology content actually attached to a Brazilian dossier, USD 1,650, with a band of USD 1,100 to USD 2,400 because it blends the minority of dossiers carrying a full battery with the majority carrying none.

That yields a 2025 band of USD 96.5 million to USD 210.5 million, and a 2033 band of USD 239.5 million to USD 358.3 million across growth rates of 6.5 and 12.0 per cent. The 2022 volume is the latest published in this form, so the base is anchored three years back and is therefore conservative, because Brazilian cosmetic regularisations have not contracted.

No second route to that number was built and presented as corroboration. Any second route available would have restated the same two unknowns — how much of a Brazilian dossier carries new testing, and what that testing costs — dressed as an independent series. One construction with its band and its falsifiers is worth more than two that agree because they share a guess.

What would prove it wrong

The first biennial report under Article 2(II) is the sharpest test. If requests for documentary evidence come back near zero, the inspection machine is not running, section 13’s exception is unlimited in practice, and this forecast collapses toward the low band.

The second is chemical rather than administrative. If OECD validation lands for repeated-dose systemic toxicity or developmental toxicity before 2030, the section 13 ceiling lifts, because the endpoints where Brazilian companies currently have no alternative become servable.

The third is logistical. The reconstructed human epidermis that carries the dermal endpoints is not made in Brazil at scale. Every assay on that tissue is a cold-chain import decision, and a domestic tissue line would not merely substitute an import — it would shorten the lead time on every dermal dossier in the country. If it does not appear before 2029, the plateau lasts longer than modelled.

The practical reading for anyone buying or selling here

The register rewards depth, not breadth. Each accreditation carries a fixed cost and a narrow revenue base, because the number of Brazilian dossiers needing any one endpoint in a year is small. A laboratory carrying six methods is not six times as profitable as one carrying two; it is six times as exposed to the cost of maintaining them. The rational strategy is the two or three methods that let a firm quote a complete dermal battery, and a competitor holding the ocular franchise instead.

The weak and valuable position is the junction between the bench and the dossier. A supplier who can deliver the assay, the documentation that defends it and the label claim it supports — on one contract and one timetable — is selling something no Brazilian laboratory currently sells. That is the plausible shape of consolidation in this market, and it is a services consolidation rather than a laboratory one.

And the register itself is still the thing to watch. Brazil’s science ministry states 46 recognised methods; adding the seven resolutions item by item gives 47, a one-method difference the published page does not resolve and which Pheonix Research flags rather than smooths. The count is a policy claim as much as a scientific one. What matters commercially is not how many are on it, but which endpoint each one covers — because a laboratory can hold a recognised method for an endpoint nobody in this market needs and be no better off for it.

The full analysis, including the competitive map, value chain, investment calendar and regulatory landscape, is in Brazil In Vitro Toxicology Testing Market Report, Size, Share and Forecast 2026-2033. Related reading: the global organ-on-chips market, LC-MS systems, histopathology and clinical laboratory services.