Latex Mattresses: What's In Them, and the Test Nobody Has Run

How a latex mattress is actually made

Latex foam starts as a liquid: the milky sap of Hevea brasiliensis, the rubber tree. To turn that liquid into a slab you can sleep on, two things have to happen. It gets whipped into a foam, and then it gets cooked with sulfur. The cooking is what turns sticky raw sap into something springy that holds its shape for years.

That sulfur cooking is called vulcanisation. It is the only piece of jargon on this page you need, and it is worth holding onto for one reason: almost everything people worry about in a latex mattress comes from the cooking step rather than from the tree.

Sulfur on its own works too slowly to be practical. So the recipe includes accelerators — chemicals whose only job is to make the sulfur cook faster — plus zinc oxide to help them along. These are the additives that matter for everything below, and they are not exotic: the common families are thiurams, mercaptobenzothiazoles and dithiocarbamates. Industry

Two commercial processes dominate. Dunlop is the older and simpler one: foam the latex, pour it into a mould, cook it. Talalay adds a vacuum and a freezing step before the cooking, which gives a lighter, more even foam. Both end up as sulfur-cooked rubber, and neither removes the need for accelerators. Industry

One thing worth knowing before you shop: "latex" on a label does not mean the slab is all tree. Latex foam is also made from synthetic styrene-butadiene rubber, and blends of the two are sold under the same word. "Latex" is not a protected term, so the wording on the tag is the manufacturer's, not an audited claim: "100% natural latex" is the phrase that excludes a blend, and it is worth getting in writing. Industry

A certificate settles it, and one does exactly this. GOLS, the Global Organic Latex Standard, states that a certified product "shall not contain synthetic latex," requires the synthetic share of polymer content to be zero, and requires that to be proven by a laboratory content-analysis test rather than declared. If you want the blend ruled out, that is the document that rules it out.

The reverse does not hold, which is where this gets misread. Plenty of latex is 100% natural and simply not organic, and it will carry no GOLS certificate at all. Presence of GOLS rules out a blend; absence of it tells you nothing. Industry

Synthetic rubber is built from two small starting chemicals, styrene and 1,3-butadiene, which get linked into long chains during manufacture. The EU has given both of them a formal hazard classification, which is the strongest kind of official listing short of a restriction. Regulatory Both also appear in the Embr Exposure Ledger: styrene in 8 of its 10 datasets, butadiene in 6. That count is our own arithmetic on the published Ledger, not a figure from a paper, and it is reproducible from the downloadable file. Inferred

What that pair of facts does not tell you is whether any of those starting chemicals survive unreacted in a finished slab, let alone whether any of it reaches you. That is a manufacturing-quality question, and I found no measurement of it in mattress foam of any kind. Inferred

"A latex mattress" is usually not a latex mattress

This is the thing that changes the answer for most people, and almost nothing written about latex says it plainly.

Very little of what is sold as a latex mattress is a solid block of latex. Far more common is latex as a comfort layer — a few inches of it quilted under the cover, sitting on a core of polyurethane foam or on pocket springs. The word on the tag describes the layer you sink into, not the thing you are lying on. Industry

Three builds get sold under the same word, and they are genuinely different products:

• All-latex. Every foam layer is latex, usually a firmer core with a softer layer above. This is the one the marketing describes and the least common at ordinary prices.
• Latex hybrid. A latex comfort layer over a pocket-spring unit. The springs are steel; the chemistry is latex plus whatever the cover and any adhesives are.
• Latex over foam. A latex comfort layer over a polyurethane core. Sold as a latex mattress, and by volume it is mostly polyurethane.

What that means for you is not subtle. If your latex sits on a polyurethane core, then the emissions literature that applies to your bed is the polyurethane literature — the chamber studies further down this page, the ones that actually exist. You did not escape that chemistry by buying latex. You put a layer of latex on top of it.

It cuts the other way for the allergen. A latex layer under a quilted cover, under a sheet, is further from your skin than a bare latex surface would be, and further than a latex pillow under a thin case. If you are allergic, that is not a safety margin worth betting on — but if you are not, it is one reason the tree-protein question stays theoretical for most sleepers. Inferred

The practical version: "is it latex?" is the wrong question to ask a salesperson. "Which layers are latex, and what is underneath them?" is the right one, and it has a specific answer that should be on the spec sheet. If nobody will tell you what the core is, that is your answer.

Does it off-gas? Nobody has published the test

This is the question that brings most people to the topic, and it is the one where the honest answer is uncomfortable.

The chamber-test literature on what mattresses emit into bedroom air is real and it is good. Two new memory foam mattresses were tracked for 32 days, with the summed VOC concentration modelled at roughly 2.7 and 4.2 µg/m³ averaged over the first year, and the long-term emission half-life on the order of 24 days. Peer-reviewed Eight polyurethane mattresses were tested under simulated sleeping conditions, where emissions rose significantly and heat turned out to be the main driver rather than humidity or carbon dioxide. Peer-reviewed

Every one of those studies is polyurethane, memory foam, polyester or textile. Searching the emissions literature for latex foam mattresses returns formulation chemistry — how to make the foam antibacterial, how to make it biodegrade, how to swap the accelerators — and no chamber test of what a finished latex mattress releases into the air of a bedroom.

So the honest position is that we do not know, and neither does anyone selling you one. A latex mattress may well emit less than polyurethane; the chemistry is simpler, and it skips the isocyanates — the reactive chemicals used to make polyurethane foam, and the most hazardous ingredients in that process (TDI and MDI). But "may well" is not a measurement, and a shop assistant telling you latex does not off-gas is not reporting a result. They are reporting an absence of results.

One thing does cut against that, and it is worth knowing because almost nobody mentions it. Certified latex has been emission-tested, even though the results are not published. GOLS requires chamber testing against fixed limits — total VOCs at 200 µg/m³ after seven days, formaldehyde at 24 µg/m³, and, uniquely for latex, nitrosamines at 0.3 µg/m³. So a GOLS certificate does mean a laboratory put that material in a chamber and it came in under those numbers. What you get is a pass, not a measurement, and only for certified product. Industry

This is the same gap the Dose Gap documents across the Atlas: presence is not dose, and an unmeasured thing is not a safe thing. It is just an unmeasured thing.

Is it healthy? There are two different allergies here

"Latex allergy" gets used for two conditions that have different causes, different mechanisms and different relevance to a mattress. Keeping them apart is most of the work.

The proteins — and this is the one that is actually about mattresses

Natural rubber carries proteins from the tree. In people who have become allergic to them, these proteins trigger the fast kind of allergic reaction — the same mechanism as a peanut or bee-sting allergy. It is the latex allergy that matters in hospitals, and it can be severe.

The question for us is narrower: are those proteins still in a finished mattress? They are. Five latex foam mattresses were ground to powder and their proteins extracted; no protein or allergen at all could be extracted from synthetic foam, while the natural-latex mattresses yielded proteins that the blood serum of latex-allergic patients recognised. The major rubber-tree allergen Hev b 1 was identified directly by amino-acid sequencing. Peer-reviewed

A later study measured how much allergen was in 22 medical and 23 everyday latex products. The test works by seeing how strongly an extract of each product blocks an antibody reaction: the more allergen the material holds, the higher the percentage. It is a ranking of allergen content, not a percentage of anything in your bedroom air.

The result worth sitting with: a latex mattress came in at 74.27%, above household rubber gloves at 49.10%, and in the same band as a toy balloon at 78.62%. Peer-reviewed

That is the strongest specific finding on this page, and it is the opposite of the marketing. The material sold as the natural, hypoallergenic option is the one carrying a measurable biological allergen — because it is natural, not despite it.

Who this actually matters for: people already sensitised to latex. That is not most people, but it is a real group, and it concentrates in healthcare workers, in people who have had many surgeries, and in children with spina bifida.

The accelerators — real allergens, wrong exposure route

The other latex allergy is delayed contact dermatitis, and it is not caused by the rubber. It is caused by the accelerators — the additives from the cooking step.

These are well characterised. A systematic review pooled 106 studies of patch-test results gathered between 1984 and 2023, and found contact allergy in 2.55% for thiuram mix, 0.83% for mercaptobenzothiazole and 3.96% for carba mix. Peer-reviewed The "mixes" there are testing kits, not mattress ingredients: a patch test applies a small blend of related accelerators at once to see which family someone reacts to.

Here is where I have to argue against my own article. Those numbers are seductive — they are large, they are recent, and they name the exact chemicals that go into latex foam. It would be easy to run them straight into a paragraph about your mattress. They will not carry that weight, for two reasons.

• Those people are not you. They are eczema patients referred to a clinic for allergy testing, not the general population. A prevalence among people who already have eczema severe enough to be investigated is not a prevalence among people shopping for a bed.
• The exposure route is gloves. Across this literature the source is overwhelmingly occupational and overwhelmingly gloves — healthcare workers, hands, wet work. I could not find a case series implicating mattresses. Skin against a fabric-covered slab for eight hours is not skin inside an occlusive glove for a shift.

So the accelerator families are genuinely sensitising, they are genuinely in the product, and the evidence that a mattress delivers enough of them to matter does not exist either way. Both halves of that sentence are true and I am not going to collapse it into a scarier one.

What the Ledger can add is regulatory standing rather than bedroom dose. Both tetramethylthiuram disulfide (CAS 137-26-8) and 2-mercaptobenzothiazole (CAS 149-30-4) carry a formal EU hazard classification, and each appears in 4 of the Ledger's 10 datasets. That is a statement about how they are regulated. It is not a statement about what reaches you through a mattress cover. Regulatory

Exposed to what, exactly?

To two things, and neither of them is a fume. That is the part worth being precise about, because "exposure" in a mattress article almost always means something you breathe, and here it does not.

The tree proteins are the exposure that has actually been measured in mattresses. The accelerators are present too, but nobody has measured what a mattress delivers of them, so they remain a known ingredient rather than a known exposure.

Hev b 1 — the shorthand for the main rubber-tree protein — and its relatives are proteins. They are not volatile, so they are not something you breathe off the slab the way you would a solvent. The route that has been demonstrated for other latex products is contact and, where a powder carries them, inhalation of that powder. A mattress under a fitted sheet has neither a powder nor bare skin against it for most people. Inferred

So the honest description is narrow: a protein that is present in the material, extractable in a lab, capable of provoking an immune response in someone already sensitised, and never measured moving from a mattress into a person. Every step of that chain is documented except the last one, which is the step that would matter to you.

How would I know if it were affecting me?

Mostly, you would already know. Latex allergy is not a vague, low-grade condition you would fail to notice — it is immediate-type, so reactions come on within minutes of contact: itching, hives or welts where skin touched the material, a runny nose or sneezing, and in serious cases wheeze or swelling. It does not present as feeling generally tired or foggy.

There is no home test, and there is no test on the mattress. Diagnosis is a clinician's job and it is a specific one — a skin-prick test, or a blood test for the antibody that drives this kind of allergy (your doctor will call it latex-specific IgE). If you suspect it, that is the conversation to have, and it is worth having before you spend money on a latex bed rather than after.

One distinction that saves people a wasted appointment. Immediate reactions on contact point at the tree proteins. A slower rash that appears a day or two later, in the shape of whatever touched you, is the other mechanism entirely — delayed contact allergy, which is about the accelerators rather than the rubber, and which is diagnosed by patch testing instead. They are different tests. Saying which pattern you have is the most useful thing you can bring to a clinician.

What about a latex pillow?

This is the better question, and I did not expect to end up saying so.

A latex pillow is the same sulfur-cooked foam as the mattress, made the same way, with the same accelerators and the same tree proteins. Nothing about the chemistry changes when the slab gets smaller. What changes is the geometry: a pillow sits directly under your face, often with a single thin case rather than a sheet plus a mattress protector, and your face is where skin is thinnest and where you are breathing.

Nobody has measured a latex pillow either. The allergen work that exists was done on mattresses, and I found no equivalent for pillows. But if you are reasoning about which piece of latex bedding puts the most material closest to you for the longest, the answer is the pillow and not the mattress, and it is also the cheaper one to replace. Inferred

Our pillow chemistry piece covers what else accumulates in one over its life, which is a separate and better-measured problem.

Does it last as long as they say?

Latex is routinely sold with a 20 to 25 year life, often against the eight-to-ten years quoted for polyurethane foam.

I went looking for the independent testing behind that and did not find it. What exists in the literature is materials science: lab measurements of how far a sample stays squashed after being compressed, and how it holds up when artificially aged, used to compare one recipe against another. Those are real tests, but they are answers to "does this recipe hold up better than that recipe," not to "how long will this slab support a body in a bedroom."

The durability claim is a manufacturer claim. It may be true. Natural rubber is genuinely springy stuff, and what owners report is the slab going soft rather than falling apart. But treat 25 years the way you would treat any number that appears only in the seller's own literature: as a marketing figure until somebody independent measures it. Industry

What this does not establish

This is the honest boundary of the page, and it is wider than I expected when I started.

• No emissions number for latex exists. I cannot tell you a latex mattress off-gasses less than memory foam, and anybody who does is extrapolating from chemistry rather than reporting a measurement.
• The mattress-allergen work is old. The direct measurements on latex mattresses date from 1999 and 2006. Manufacturing has moved since. Latex with the tree proteins stripped out now exists, and so do accelerator recipes designed to avoid forming nitrosamines, a group of carcinogens that sulfur-cured rubber can generate. Whether a mattress bought in 2026 tests like a mattress bought in 1999 is unknown, and that cuts in the industry's favour, not against it.
• Nothing here is a dose. Every number on this page is a presence or a prevalence. None of them tells you how much of anything reaches a sleeping person, which is the number that would actually settle the question.
• Sensitisation risk is not quantified for bedding. That proteins are present and extractable does not establish that sleeping on them sensitises a previously unsensitised person. The 1999 authors said "potential source" and "at least to allergic individuals," and that hedge is theirs, not mine.

What I would actually do

If you have a diagnosed latex allergy: don't. This is the one clear call on the page. The proteins are in the product, they are extractable, and a barrier cover is an untested mitigation rather than a solution.

If you're buying latex because you were told it doesn't off-gas: buy it for other reasons. It may be true; it is not established. Buy it because you like how it sleeps, because it is a simpler material with fewer additives than a polyurethane build, or because you want a mattress that will biodegrade. Those are defensible. "It doesn't off-gas" is not yet.

If you're buying anyway: ask for 100% natural latex in writing rather than "natural latex," which permits a synthetic blend. Ask which certification covers the latex core specifically: GOLS bans a synthetic blend and requires emission testing, and it is the one certificate that answers the blend question outright. And air it hard for the first fortnight regardless of material — heat drives emissions from every foam that has been tested, so a warm, well-ventilated room does more than a cool sealed one.

The thing I would not do is treat latex as the safe default. It is a different set of trade-offs from polyurethane, not an absence of them. On one axis — a measurable biological allergen sitting in the finished product — it is the worse choice, and it is the only mattress material where that is true.

Frequently asked questions

References

1. Chardin H, Desvaux FX, Mayer C, Sénéchal H, Peltre G. Protein and allergen analysis of latex mattresses. International Archives of Allergy and Immunology. 1999;119(3):239-46. doi:10.1159/000024200 (PMID 10436396) Peer-reviewed
2. Crippa M, Belleri L, Mistrello G, Tedoldi C, Alessio L. Prevention of latex allergy among health care workers and in the general population: latex protein content in devices commonly used in hospitals and general practice. International Archives of Occupational and Environmental Health. 2006;79(7):550-7. doi:10.1007/s00420-005-0080-5 (PMID 16468057) Peer-reviewed
3. Isufi D, Kursawe Larsen C, Jensen MB, et al. Prevalence of contact allergy to rubber accelerators from the European baseline series in dermatitis patients: a systematic review and meta-analysis. Contact Dermatitis. 2025;93(6):439-463. doi:10.1111/cod.70024 (PMID 40922039) Peer-reviewed
4. Beckett EM, Miller E, Unice K, Russman E, Pierce JS. Evaluation of volatile organic compound (VOC) emissions from memory foam mattresses and potential implications for consumer health risk. Chemosphere. 2022;303(Pt 1):134945. doi:10.1016/j.chemosphere.2022.134945 (PMID 35588879) Peer-reviewed
5. Oz K, Merav B, Fishbain B, Broday DM, Dubowski Y. Volatile organic compound emissions from polyurethane mattresses under variable environmental conditions. Environmental Science & Technology. 2019. doi:10.1021/acs.est.9b01557 (PMID 31290311) Peer-reviewed