In this article
The short answer
Section 8 of Part A is the toxicological profile: for every substance in the product, the assessor records what is known about each relevant endpoint (irritation, sensitisation, absorption, repeated dose toxicity, genotoxicity, reproductive toxicity, phototoxicity and others), with a source for each conclusion. It produces the point of departure, usually a NOAEL or BMDL, that feeds the margin of safety in Part B. Data come from SCCS and other committee opinions, published reviews, studies and supplier dossiers. Because animal testing for cosmetics is banned, gaps are filled with existing data, validated non animal methods, justified read-across or the threshold of toxicological concern. If none of those works, the ingredient is restricted or removed.
Key takeaways
- Annex I requires a profile for every substance, covering all relevant endpoints, with particular focus on irritation, sensitisation and, for UV absorbers, phototoxicity.
- The assessor chooses which endpoints are relevant, and has to justify in writing any that are left out.
- The point of departure (NOAEL, or BMDL where possible) and any extra assessment factors applied to it set how much of an ingredient the margin of safety allows.
- Read-across must be substantiated, and TTC covers systemic effects only, with a much lower threshold for anything with a genotoxicity alert.
- Impurities, particle size and interactions between ingredients must be considered, not just the ingredients themselves.
- The documents you send, especially IFRA certificates, allergen declarations, GC analyses and supplier toxicology dossiers, decide how strong section 8 can be.
Here is a folder we see all the time. A new brand sends its serum formula with neat supplier paperwork, and for the hero peptide there is a single page: the supplier's logo, a recommended use level of up to 5%, and the sentence "non toxic and safe for cosmetic use." No studies, no absorption data, nothing an assessor can put a number on. Then come the emails to the supplier, a wait for a real toxicology summary, and quite often a supportable level well below the one on the marketing sheet. That gap between what a supplier says and what a safety assessor can use is what section 8 of Part A is all about.
If Part B is the safety assessor's verdict, section 8 is the evidence behind it: the toxicological profile of every substance in the product. It is usually the longest part of a CPSR and the hardest for a brand owner to read. You do not need to be a toxicologist to understand it, though, and knowing how it is built will save you time, money and the occasional unpleasant surprise in Part B.
What Annex I actually asks for
The requirement comes from point 8 of Part A of Annex I to the Cosmetics Regulation. Stripped of legal phrasing, it says the report must contain:
- the toxicological profile of each substance for all relevant toxicological endpoints, with particular focus on local effects (skin and eye irritation), skin sensitisation and, for anything that absorbs UV, photo induced toxicity;
- consideration of all significant routes of absorption and of systemic effects, with a margin of safety based on a no observed adverse effect level (NOAEL), or a justification if this is not done;
- particular consideration of particle size (including nanomaterials), impurities in substances and raw materials, and interactions between substances;
- a substantiated justification for any read-across; and
- the source of every piece of information, clearly identified.
The Commission's guidelines on Annex I (Implementing Decision 2013/674/EU) fill in the detail. They list the endpoints that may be relevant, rank the types of data from guideline studies down to screening data, and make one point brand owners often miss: choosing which endpoints matter is the safety assessor's responsibility, and any endpoint judged irrelevant has to be justified in writing.
In Great Britain the same Annex I text applies through the Cosmetics Regulation as retained and amended after Brexit, and it is enforced under the Cosmetic Products Enforcement Regulations 2013, mostly by Trading Standards. In the EU, and in Northern Ireland, it is Regulation (EC) No 1223/2009 itself. The science expected in section 8 is the same in both.
The endpoints an assessor works through
For each substance, the profile is organised by endpoint, which is just the toxicologist's word for a type of harm. The guidelines list twelve. Not every one needs fresh data for every ingredient: shea butter does not need a carcinogenicity discussion running to pages. What matters is that each one is considered, and that the assessor writes down why an endpoint does or does not apply.
| Endpoint | The question it answers | Where it usually shows up in Part B |
|---|---|---|
| Acute toxicity (oral, dermal, inhalation) | What happens after one large exposure, such as a child drinking bath foam | Accidental ingestion, sprays, child resistant packaging decisions |
| Skin irritation and corrosion | Does it damage or inflame skin, and from what level? | Concentration caps, leave on versus rinse off decisions, warnings |
| Eye irritation (mucous membranes) | What happens if it gets in the eye? | Eye warnings, products used near the eye |
| Skin sensitisation | Can it cause contact allergy, and how potent is it? | Fragrance and preservative levels, restrictions for sensitive users |
| Dermal or percutaneous absorption | How much crosses the skin? | The systemic exposure dose, and so the margin of safety |
| Repeated dose toxicity (usually 28 or 90 day studies) | What happens with daily exposure, and at what dose does nothing adverse happen? | The point of departure for the margin of safety |
| Mutagenicity and genotoxicity | Can it damage DNA? | Whether a threshold approach is allowed at all |
| Carcinogenicity | Can it cause cancer, and by what mechanism? | Acceptability, links to Article 15 CMR rules |
| Reproductive toxicity | Effects on fertility and on the developing child | Products used in pregnancy, CMR status |
| Toxicokinetics (ADME) | How is it absorbed, distributed, metabolised and excreted? | Route to route corrections, refining absorption |
| Photo induced toxicity | Does UV light make it irritating, phototoxic or photoallergenic? | Leave on products with UV absorbing ingredients or citrus oils |
| Human data | Patch tests, clinical use, complaints, post market surveillance | Often decisive for irritation and allergy in real use |
Two corrections to things we often see in older reports. First, sensitisation data tell the assessor how much of a sensitiser is safe in your product type. They do not decide whether a fragrance allergen is named on your label. That is a separate labelling rule with fixed thresholds in Annex III (0.001% in leave on and 0.01% in rinse off products), explained in our guide to fragrance allergen labelling. Second, a single positive genotoxicity result in a test tube does not automatically sink an ingredient. The SCCS Notes of Guidance allow follow up testing and a weight of evidence review. Only if a clear positive cannot be overruled is the substance treated as a mutagen, and at that point no safe threshold is assumed.
The number that leaves section 8
Most of a profile is descriptive, but one figure travels on into Part B: the point of departure for systemic toxicity. Usually that is the NOAEL from the best repeated dose study, expressed in mg per kg of body weight per day. The SCCS Notes of Guidance, 12th revision now say the preferred method, where good in vivo data allow it, is a benchmark dose (BMDL), with a NOAEL or LOAEL used where a BMDL cannot be calculated.
The study behind the number matters as much as the number. If the only study is 28 days rather than 90, the SCCS commonly applies an extra assessment factor of 3. If the study found effects at every dose, so only a LOAEL exists, a further factor of 3, and sometimes up to 10, is usually added. Section 8 should state which study was used and which factors were applied, because those choices can easily halve or double the level of an ingredient you are allowed.
That point of departure is then divided by the systemic exposure dose worked out in sections 6 and 7 to give the margin of safety, which is normally expected to be at least 100. We cover that calculation in the margin of safety, explained simply, and the exposure side in our guide to exposure assessment (Part A, sections 5 to 7), so we will not repeat it here.
Where the data come from
A good assessor does not start from scratch. They work down a rough order of reliability, and cite whatever they use:
- Opinions of official scientific committees. SCCS opinions cover preservatives, UV filters, hair dyes, colourants, nanomaterials and ingredients of concern, and usually name the point of departure the committee accepted. In Great Britain, the Scientific Advisory Group on Chemical Safety (SAG-CS) now advises the Office for Product Safety and Standards in the same way. One caution: an opinion is advice. It changes the law only when the annexes are amended, and the GB and EU annexes no longer move together.
- Other authoritative reviews. Cosmetic Ingredient Review safety assessments from the US, registration dossiers held by the European Chemicals Agency, evaluations by food safety bodies for ingredients that are also food additives, and the Research Institute for Fragrance Materials for fragrance materials.
- Peer reviewed studies, weighed for design, guideline compliance and relevance to skin exposure.
- Supplier dossiers and toxicology summaries. Often essential for newer ingredients, but read with care: the supplier wants the ingredient sold.
- Screening data such as in silico predictions and read-across, used to support, not replace, real data where they exist.
When studies disagree, which happens more than people think, the assessor uses a weight of evidence approach. A well run guideline study outweighs a poorly reported one, and a committee's conclusion, having already done that weighing, generally carries the most weight. The profile should say which data were preferred and why, so an inspector can follow the reasoning rather than take it on trust.
Working within the animal testing ban
Article 18 bans animal testing of finished cosmetics and of their ingredients to meet the cosmetics rules, and bans selling products where such testing took place. In the EU the testing ban has applied to finished products since 11 September 2004 and to ingredients since 11 March 2009. The marketing ban followed in March 2009 for most endpoints, and from 11 March 2013 for the last three: repeated dose toxicity, reproductive toxicity and toxicokinetics. The GB text keeps the same bans and expressly allows historic animal data to be used.
In practice, section 8 leans on three things. Existing data, including studies done before the bans or for other legal purposes. Validated non animal methods, for example reconstructed human epidermis for skin irritation (OECD TG 439), reconstructed cornea models for eye irritation (TG 492), and the in chemico and in vitro sensitisation assays (TG 442C, 442D and 442E) combined in the defined approaches of TG 497. And human data, such as patch testing that confirms tolerance at the intended level once safety is already established. What you will not see in a credible report is a new animal study commissioned for your cosmetic.
A quick test of your own report
Pick your two least familiar ingredients and find them in section 8. If you can see an endpoint list, a named source for each conclusion and a point of departure with its study, the profile has been built properly. If all you find is a sentence saying the ingredient is commonly used, ask your assessor to explain.
When the data run out: read-across, TTC or exclusion
Plenty of cosmetic ingredients have patchy data. The assessor then has three honest options.
Read-across. Data are borrowed from one or more closely similar substances. A typical example is an ester with no repeated dose study of its own, where close relatives break down in the body to the same alcohol and fatty acid and have been well studied. The SCCS expects similarity to be shown not just in structure but in physicochemical behaviour, metabolism and effects, and Annex I requires every read-across to be substantiated and justified. A line saying "read across from similar esters" with no named analogue is not enough.
Threshold of toxicological concern (TTC). For substances present at very low levels whose structure and exposure are known, the SCCS accepts exposure thresholds below which systemic harm is very unlikely. Its currently recommended values are 46 µg per kg body weight per day for Cramer class I (simple, readily metabolised structures) and 2.3 µg/kg bw/day for class III, with class II treated as class III. For anything with a structural alert for DNA reactivity, the threshold drops to 0.0025 µg/kg bw/day.
To see what that means, take an unidentified trace constituent at 1 ppm in a body lotion. Using the SCCS figure of 123.2 mg of body lotion per kg body weight per day and a 50% absorption default, the systemic dose is about 0.06 µg/kg bw/day. That sits comfortably under the class III threshold, but it is about 25 times the threshold for a substance with a genotoxicity alert. Same trace, same product, very different answer.
TTC has firm limits. It covers systemic toxicity only, so it says nothing about sensitisation or irritation. The SCCS points to EFSA's exclusion list: high potency carcinogens, metals and inorganics, proteins, steroids, bioaccumulating substances, nanomaterials and mixtures of unknown structure are all out of scope. It also cannot stand alone for substances regulated in the annexes, such as preservatives or UV filters.
Exclusion. If there are no usable data, no defensible analogue and the exposure is above any threshold, the assessor cannot conclude the ingredient is safe and will ask for it to be removed, replaced or used at a level the data do support. That is a finding about the evidence, not a verdict on the ingredient, and better data can change it.
Impurities, particle size and interactions
These three are named in Annex I, and a profile that ignores them is incomplete. Impurities include 1,4-dioxane in ethoxylated surfactants, heavy metals in clays and mineral pigments, and the nitrosamine limits Annex III sets for certain amine ingredients. Article 17 tolerates technically unavoidable traces of prohibited substances only if the product stays safe, and the profile is where that is shown.
Particle size matters because a nanomaterial can behave differently from the bulk form of the same substance, so data on the bulk material may not cover it. Interactions are about combinations: a secondary amine sitting alongside a nitrosating agent can form nitrosamines in the product, which is why the assessor looks at the formula as a whole and not just the ingredient list one line at a time. For CMR substances, which have their own rules under Article 15, see our article on CMR reclassification.
Fragrance and essential oils: one line on your formula, dozens in the profile
A fragrance listed as "parfum" at 1% might contain thirty or more materials. Section 8 has to deal with the ones that matter toxicologically: the sensitisers, the phototoxic materials, anything restricted in Annex III, and any constituent present above a level a threshold approach can cover. The assessor builds this from the IFRA certificate, the allergen declaration and, where the supplier will share it, a fuller breakdown. Essential oils are handled the same way, constituent by constituent, and are often harder because composition shifts from batch to batch. Bergamot, lime and other cold pressed citrus oils need a phototoxicity look in any leave on product, which our guide to phototoxic essential oils covers in detail. For how a toxicologist works through a perfume oil specifically, see toxicological risk assessment for fragrances and perfume oils.
What to send your assessor, ingredient by ingredient
This is where you have real influence. The strength of section 8 depends heavily on the documents you hand over, and the same assessor can do far more with a complete pack than with a folder of marketing sheets.
| Ingredient type | What the assessor needs to build its profile | What usually goes wrong |
|---|---|---|
| Plant oils and butters (sunflower, shea, jojoba) | INCI, CAS, certificate of analysis or specification, refined or unrefined, supplier SDS | Little, as long as the specification is current. Peroxide value matters mainly for stability and rancidity |
| Fragrance oils | IFRA certificate for the current amendment, allergen declaration covering the expanded EU list, SDS, ideally a full or partial constituent breakdown | Old IFRA certificates, allergen lists that stop at the original 26, no information on non allergen constituents of concern |
| Essential oils | Botanical name and plant part, GC analysis for the batch or a typical batch, allergen declaration, IFRA certificate | Composition varies by season and source, so a generic data sheet hides real differences |
| Botanical extracts | Plant part, extraction solvent, drug to extract ratio, carrier and preservative in the extract, any toxicology on the extract itself | The extract is assessed as if it were the whole plant, or the carrier (often glycerin or alcohol) and its preservative are missed |
| Actives, peptides and trend ingredients | Supplier toxicology dossier, dermal absorption data, purity and impurity profile, recommended use level with the basis for it | A one page "safe and non toxic" statement with no studies behind it |
| Preservatives | Exact level, the Annex V entry, any SCCS opinion, and how preservative also arrives through other raw materials | Preservative hidden in an extract or a fragrance pushes the total above what was assessed |
| Colourants, micas and pigments | CI number, Annex IV entry and field of application, heavy metal specification | A pigment approved for rinse off or non mucous membrane use goes into a lip product |
| Mineral and nano grade materials | Particle size distribution, whether the material meets the nanomaterial definition, coating, form | A "non nano" claim with no particle size data |
Our CPSR document checklist covers the rest of Part A, and IFRA Standards and certificates explained shows what a current certificate should contain.
What a strong section 8 looks like
| A strong section 8 | A weak one |
|---|---|
| A profile for each ingredient, organised by endpoint | One generic paragraph per ingredient, or ingredients lumped together with no reason given |
| Every data point cited: the SCCS opinion, the CIR report, the study | No sources, or "widely used and considered safe" |
| The point of departure used in the margin of safety is named, with its study and any extra assessment factor | Margins of safety appear in Part B with no visible origin |
| Relevant fragrance and essential oil constituents profiled individually | The fragrance treated as one ingredient |
| Read-across and TTC explained and justified where used | Data gaps quietly skipped |
| Endpoints left out are named, with the reason | Endpoints simply missing |
| Impurities, particle size and interactions considered | No mention of any of them |
Length on its own proves nothing, but a section 8 that runs to a single page for a twenty ingredient product has not done the work. It will not stand up to a Trading Standards officer asking how the margin of safety was reached, and it gives your insurer little to rely on if a customer reacts.
Ingredients that take the longest
- Essential oils and complex botanicals: variable composition, frequent sensitisers, occasional phototoxins, and thin data on the whole material.
- New actives and trend ingredients: peptides, ferments, exotic extracts. Often supplier data only, and sometimes not enough.
- Preservatives: well studied, but with annex maximums, committee opinions and total exposure from other products to apply precisely.
- Colourants: purity, heavy metals and the permitted field of application.
- Anything in a lip, eye or baby product: ingestion, mucous membranes or a child's higher exposure per kilogram raise the bar for every endpoint. See CPSR for products used on under threes.
Want a section 8 you can defend?
Send us your formula and the supplier documents you already hold. We will tell you which ingredients are well covered, which need more data before assessment, and which are likely to be restricted, so you can fix the problem materials before you commit to a batch.
Related reading
- CPSR Part A explained: all ten elements
- The margin of safety, explained simply
- Exposure assessment in a CPSR (Part A, sections 5 to 7)
- Why your CPSR came back with conditions
- The documents your assessor needs for Part A
- IFRA Standards and certificates explained
- Phototoxic essential oils in cosmetics
- CMR reclassification and your CPSR
- All regulatory compliance articles
Myths we hear
- If an ingredient is natural, it does not need a toxicological profile. Essential oils and plant extracts often need the most work, because they contain many constituents, including sensitisers and phototoxins.
- A supplier statement that the ingredient is safe is enough. An assessor needs data they can cite: studies, committee opinions or a toxicology dossier. A marketing claim is not evidence.
- No animal testing means there is no safety data. Assessors use existing studies, validated non animal methods, human data, read-across and TTC. The ban changed the tools, not the standard.
- Sensitisation data decide which allergens go on my label. Allergen labelling follows fixed thresholds in Annex III. Sensitisation data decide how much is safe.
- An SCCS opinion is the law. It is scientific advice. The legal limit changes only when the annexes are amended, separately in the EU and in Great Britain.
What to do this week
- List every raw material and mark which ones you have a specification, SDS and certificate of analysis for.
- Check that each fragrance has an IFRA certificate for the current amendment and an allergen declaration covering the expanded EU list.
- Ask the supplier of any active or new ingredient for its toxicology dossier and dermal absorption data, not just the marketing sheet.
- For essential oils and botanical extracts, request a GC analysis, the plant part, the extraction solvent and the carrier.
- Open your current CPSR, find your two least familiar ingredients in section 8, and check that each conclusion has a named source.
Frequently asked questions
What is the toxicological profile in a CPSR?
It is point 8 of Part A: a record, for each substance in the product, of what is known about its toxic effects across the relevant endpoints, with the source of each piece of information. It is the evidence behind the margin of safety, the warnings and any restrictions in Part B.
Which endpoints does the profile cover?
The Commission guidelines list acute toxicity, irritation and corrosion, eye irritation, skin sensitisation, dermal absorption, repeated dose toxicity, genotoxicity, carcinogenicity, reproductive toxicity, toxicokinetics and photo induced toxicity, plus human data. The assessor decides which are relevant to each ingredient and justifies any left out.
Where does the NOAEL in my report come from?
Usually from the best repeated dose study for the ingredient, often taken from an SCCS opinion or another authoritative review. The SCCS now prefers a benchmark dose (BMDL) where the data allow. Section 8 should name the study and any extra factors, for example 3 for a 28 day study instead of a 90 day one.
How can an ingredient be assessed without animal testing?
The assessor uses existing data, including studies done before the bans, validated non animal methods such as reconstructed skin models and in vitro sensitisation assays, human data, and justified read-across or TTC. New animal tests to meet the cosmetics rules are banned in the EU and Great Britain.
What is read-across?
Using data from closely similar substances to fill a gap for an ingredient that has not been tested for a given endpoint. Annex I requires every read-across to be substantiated and justified, so the report should name the analogue and explain why it behaves the same way.
When can the threshold of toxicological concern be used?
For substances at very low levels whose structure and exposure are known, such as trace impurities or minor constituents. It covers systemic toxicity only, uses a far lower threshold for anything with a genotoxicity alert, excludes groups such as metals, proteins and nanomaterials, and cannot be used alone for annex regulated ingredients.
Why does my assessor keep asking my supplier for more data?
Because a marketing sheet or a one line safety statement cannot be cited. For actives and new ingredients the assessor needs studies or a toxicology dossier, ideally with dermal absorption data, which can also support a higher use level than a cautious default would allow.
Does a fragrance count as one ingredient in section 8?
No. The assessor profiles the constituents that matter toxicologically, such as sensitisers, phototoxic materials and restricted substances, using the IFRA certificate, the allergen declaration and any fuller breakdown the supplier provides.
Official sources and legislation
- GB Cosmetics Regulation, Annex I: cosmetic product safety report, Part A point 8 (legislation.gov.uk)
- GB Cosmetics Regulation, Article 18: animal testing
- GB Cosmetics Regulation, Article 15: CMR substances
- Regulation (EC) No 1223/2009, EU text as adopted (legislation.gov.uk)
- Commission Implementing Decision 2013/674/EU: guidelines on Annex I
- SCCS Notes of Guidance for the testing of cosmetic ingredients and their safety evaluation, 12th revision
- SCCS opinions (European Commission)
- Scientific Advisory Group on Chemical Safety in Consumer Products (SAG-CS), GOV.UK
- Cosmetic Ingredient Review: ingredient safety assessments
- European Commission: ban on animal testing for cosmetics
- IFRA Standards library (International Fragrance Association)
- Cosmetic Products Enforcement Regulations 2013 (SI 2013/1478)
- OPSS: making cosmetic products available to consumers in Great Britain
Join the conversation