Rosemary extract as an antioxidant: dose, carnosic acid and honest limits
Rosemary extract at 0.02 to 0.5 percent depending on carnosic acid content. Induction time data, the colour and smell it brings, and why it is not a preservative.
A percentage on a rosemary extract bottle means almost nothing on its own. Dose 0.1 percent of a 5 percent extract and 0.1 percent of a 25 percent extract and you have put fivefold different amounts of the active into your oil. This page works the dose backwards from carnosic acid, gives the oxidation stability numbers that justify it, and is specific about what the material will not do.
Dose on carnosic acid, not on the bottle. Aim for 50 to 250 ppm of carnosic acid plus carnosol in the finished balm. That is 0.2 to 0.5 percent of a 5 percent extract, 0.1 to 0.25 percent of a 10 percent grade, or 0.02 to 0.1 percent of a 25 percent grade. Buy oil soluble and deodorised.
What is actually in the bottle
Rosemary extract, often sold as ROE or rosemary oleoresin extract, is a concentrate from the leaves of Rosmarinus officinalis. Botanists have moved the species into Salvia, so Salvia rosmarinus appears on newer certificates for the same plant, while the cosmetic naming convention has kept the old genus.
Two diterpenes carry nearly all of the antioxidant activity: carnosic acid and carnosol. Carnosic acid is an ortho-diphenol, which is what makes it a fast hydrogen donor to lipid radicals. Carnosol is one of its own oxidation products, still active, and rosmanol and epirosmanol sit further down the same degradation chain. This is why assays and food law both specify the sum of carnosic acid and carnosol rather than either alone: the ratio between them tells you how much of the material has already been spent, and a batch high in carnosol relative to carnosic acid is an older or harder-processed one.
Rosmarinic acid is the other name you will see on rosemary literature. It is a water soluble phenolic acid, it dominates aqueous and hydroalcoholic extracts, and it is largely irrelevant in a balm because it does not partition into the oil phase. If a certificate quotes only rosmarinic acid content, you are looking at an extract designed for a water phase, not for protecting oils. The relevance of solvent match is covered under solubility in anhydrous formulas.
Everything else in the bottle is carrier. Liquid grades are usually the active dissolved or dispersed in sunflower oil or in caprylic/capric triglyceride, which is why a 5 percent grade is mostly oil, and why the carrier gets its own entry on your ingredient list.
Rosemary extract is not rosemary essential oil
These are two different materials with two different INCI names, and the confusion costs people their shelf life. Rosmarinus Officinalis Leaf Oil is the steam distilled essential oil: 1,8-cineole, camphor, alpha-pinene, borneol and camphene. Rosmarinus Officinalis Leaf Extract is the solvent or supercritical extract that carries the diterpenes.
Carnosic acid is a non-volatile molecule with a molecular weight above 330. It does not come over in steam distillation, so a rosemary essential oil contains essentially none of it. Buying essential oil as an antioxidant delivers the smell of the herb and negligible protection, plus a camphor and cineole load that a lip product does not want. Essential oil in a balm is a fragrance decision with its own limits, set out in essential oils in balms and, for the camphor question specifically, in menthol, camphor and actives.
A thin, water-clear liquid that smells sharply of the herb garden and pours like water is an essential oil. A genuine oleoresin extract is viscous, ranges from pale amber to dark green-brown, and smells resinous rather than fresh. If the supplier will not state carnosic acid content, assume there is none to state.
Dosing backwards from carnosic acid
Decide the concentration of active you want in the finished product first, then divide by the assay on the certificate. A defensible target band for a balm is 50 to 250 ppm of carnosic acid plus carnosol, which is 0.005 to 0.025 percent by weight of the finished formula. The lower end is roughly where measurable protection begins in bulk oil work; the upper end sits at about the maximum permitted level for the same actives in food fats, so it is a ceiling with a regulator's arithmetic behind it rather than a guess.
| Assay (carnosic acid + carnosol) | Dose for 100 ppm | Dose for 250 ppm | Grams per 250 g at 100 ppm |
|---|---|---|---|
| 5% (standard food grade minimum) | 0.20% | 0.50% | 0.50 g |
| 10% (common deodorised liquid) | 0.10% | 0.25% | 0.25 g |
| 25% (concentrated, often decolourised) | 0.04% | 0.10% | 0.10 g |
| 60% (purified carnosic acid fraction) | 0.017% | 0.042% | 0.04 g |
That table is the whole of the 0.02 to 0.5 percent range quoted for rosemary extract. It is not a range of strengths, it is one target expressed through four different products. A maker who reads "use at 0.5 percent" and applies it to a 25 percent concentrate has dosed at over 1,200 ppm, five times the food ceiling, and will be able to smell and taste the result.
Two refinements. First, a balm is not all oil: at 15 percent wax the oil phase is 85 percent of the mass, so 100 ppm on the whole formula is about 118 ppm on the oil that actually oxidises. Dose on the total formula anyway, because that is how every other percentage in your worksheet is expressed, and treat the difference as headroom. Second, unrefined and infused oils arrive with their own phenolics and their own metal load, so an infused oil is a different starting point from a refined one.
0.04 g cannot be weighed on a scale that reads to 0.01 g with any confidence. Make a 1 in 10 predilution: 1 g of extract into 9 g of the same carrier oil you are formulating with, kept in a small dark bottle, then weigh 0.4 g of the dilution. The batch calculator handles the arithmetic for other batch sizes, and weighing and calibration covers what your scale can honestly resolve.
What the induction time data actually shows
The instrument behind almost every antioxidant claim is the Rancimat or the equivalent oil stability index apparatus, AOCS Official Method Cd 12b-92. Air is bubbled through oil held at 110 or 120 C, the volatile acids produced when the oil breaks down are swept into water, and the conductivity of that water is logged. The induction period is the breakpoint where conductivity turns sharply upward, and it is reported in hours.
| Oil | No additive | Rosemary extract 400 mg/kg | Fold gain |
|---|---|---|---|
| Soybean | 2.20 h | 3.40 h | 1.5x |
| Rapeseed | 2.32 h | 4.15 h | 1.8x |
| Cottonseed | 1.88 h | 3.35 h | 1.8x |
| Rice bran | 3.83 h | 6.22 h | 1.6x |
| Camellia | 1.54 h | 3.39 h | 2.2x |
Read that as a fold gain of roughly 1.5 to 2.2 at a dose of 400 mg/kg of extract, not of active. The consistent finding across bulk oil studies is that the response is dose dependent over the practical range and that the gain is larger in oils that start weakest, which is the pattern in the camellia and cottonseed rows above.
Against tocopherol alone the comparison is less flattering to tocopherol than makers expect, for a reason that has nothing to do with rosemary. Most seed oils already contain tocopherols: sunflower carries 400 to 935 mg/kg of alpha-tocopherol, which is four to nine times the concentration measured as optimal in bulk oil, so adding more frequently produces a flat or negative induction time response. Rosemary extract is being added to an oil that contains none of it, and the marginal return is therefore much higher. Neither material is "stronger" in the abstract, and the head-to-head is worked through in vitamin E versus rosemary extract.
Do not convert induction hours into shelf months. The usual rule of thumb doubles reaction time for every 10 C drop, so scaling a 120 C result to a 20 C shelf multiplies by roughly a thousand, and every error in the assumption multiplies with it. The ranking transfers; the absolute prediction does not. See accelerated ageing tests.
Crude, deodorised and decolourised grades
Rosemary extract is produced by extraction with acetone, ethanol or supercritical carbon dioxide, followed by optional purification steps. Deodorising is usually molecular distillation or steam stripping, which removes the volatile terpenes that carry the smell. Decolourising uses an adsorbent step to strip the chlorophyll and carotenoid pigments. The diterpene actives are non-volatile and survive both in principle, but every extra process step costs yield, which is why the purest-looking grades are also the dearest.
| Grade | Typical assay | Colour it adds | Smell and taste | Where it belongs |
|---|---|---|---|---|
| Crude oleoresin | 5-10% | Dark green to green-brown, visible at 0.2% | Strong resinous herbal, obvious on the lip | Dark, heavily scented or bulk oil use only. Cheapest per unit of active |
| Deodorised liquid | 5-15% | Amber to olive, a faint cast at 0.1% | Slight herbal note, usually masked by fragrance | The sensible default for tins, salves and scented balms |
| Deodorised and decolourised | 10-25% | Pale straw to near colourless | Effectively neutral at in-use levels | Pale lip balms, unscented ranges, delicate fragrances |
| Powder on a solid carrier | 3-10% | Off-white to buff particles | Neutral | Dry blends and water phases. Not for a balm |
The sensory cost is the real constraint, not the price. A crude oleoresin at 0.4 percent in an unflavoured lip balm is detectable on the lip within a second, and it reads as bitter rather than herbal. In a tinted or naturally coloured product the green cast also shifts the shade: an ivory base turns faintly olive, and a pink pigment goes muddy. If a batch has landed with a taste you did not intend, lip balm tastes soapy works through the usual suspects, and balm smells too strong covers the aroma side.
Oil soluble against water dispersible
Carnosic acid is strongly lipophilic and dissolves readily in triglyceride oil. That is the grade a balm needs, supplied either as a liquid in sunflower oil or MCT, or as a soft solid that goes into warm oil without complaint. Stir it into the oil phase and give it time to disperse fully; an antioxidant sitting in one corner of the pot protects one corner of the pot.
Water dispersible grades exist because most food uses are emulsions or aqueous systems. They are made by adsorbing the extract onto maltodextrin or silica, or by pairing it with a solubiliser such as polysorbate 80. In an anhydrous balm neither route dissolves. The maltodextrin version leaves particles that read as grit under the finger and can look like the fault described in undissolved wax specks, and the solubilised version drags on application. Check the technical data sheet for the word "dispersible" before you buy, because craft suppliers rarely put it in the product name.
One subtlety worth knowing, because it explains contradictory claims. In bulk oil, more polar antioxidants tend to outperform less polar ones, while in an oil-in-water emulsion the order flips. A balm is the bulk oil case, so data generated in emulsions or in food products with a water phase does not transfer cleanly. Rosemary extract is a mixture spanning a range of polarities, which is part of why published results for it vary so much between systems.
It protects the oils, not the product
Rosemary extract slows the chemical reaction between oxygen and unsaturated fatty acids. It does nothing useful against bacteria, yeast or mould at the concentrations a balm can carry. These are unrelated failure modes with unrelated countermeasures, laid out in rancidity and oxidation and do balms need preservatives.
The antimicrobial confusion has a real root. Carnosic acid does show activity against Gram-positive bacteria in laboratory assays, which is why "natural preservative" copy attaches itself to rosemary extract. The minimum inhibitory concentrations reported for those assays sit in the hundreds of milligrams per litre, orders of magnitude above the 50 to 250 ppm this page recommends, and the activity does not extend usefully to Gram-negatives, yeasts or moulds. Rosemary extract is not listed in Annex V of Regulation (EC) No 1223/2009, the closed list of preservatives permitted in EU cosmetics, and a product containing water and dosed with it will fail an ISO 11930 challenge test. If your formula contains a hydrosol, honey, aloe or fresh plant material, or will be scooped with wet fingers, it needs a real preservative system: see preservatives for anhydrous products and mould in balm.
It also cannot reverse damage. Once hydroperoxides have decomposed into the aldehydes you smell as crayon or old paint, those are stable end products and there is nothing left for a radical scavenger to act on. Adding extract to an oil that already smells wrong dilutes the smell and does nothing else, which is the first point made in rancid balm.
Names, E 392 and how it appears on a label
On a cosmetic ingredient list the material is Rosmarinus Officinalis (Rosemary) Leaf Extract, and the carrier is declared separately, usually as Helianthus Annuus (Sunflower) Seed Oil or Caprylic/Capric Triglyceride. Because it is used at well under 1 percent it can be placed anywhere after the last ingredient present above 1 percent, so its position on a label tells you nothing about the dose. "ROE" is trade shorthand and is not an INCI name. How the rest of the list is read is covered in reading a balm label, and the INCI list builder puts a list in the right order.
In food it is E 392, extracts of rosemary. Commission Regulation (EU) No 231/2012 defines the permitted production processes and requires the specified carnosic acid and carnosol content; Annex II of Regulation (EC) No 1333/2008 sets maximum levels across food categories, expressed as the sum of the two actives on the fat content, running from 15 up to 400 mg/kg. EFSA's re-evaluation derived an acceptable daily intake of 0.3 mg per kg of body weight per day. In the United States, rosemary in the form of spice, essential oil and oleoresin is listed as generally recognised as safe under 21 CFR 182.20, and extracts sold specifically as antioxidants are marketed on GRAS determinations.
None of that is a cosmetic authorisation, and none of it entitles you to call the ingredient a preservative on a label or in copy. In a lip balm the food figures matter for a different reason: lip products are ingested in small amounts, so the food ceiling is a sensible sanity check on your dose rather than a legal requirement. Claim language generally is covered in cosmetic versus drug claims.
Combining with tocopherol and a chelator
The blend that measures best in food laboratories is rosemary extract plus tocopherols plus a chelator, usually citric acid, sometimes with ascorbyl palmitate. Two of those three travel badly into a balm.
The tocopherol pairing is genuinely disputed rather than settled. One line of work reports an optimum near a 1.5 to 1 ratio of tocopherol to carnosic acid plus carnosol; Frankel and colleagues, working in bulk corn oil, found alpha-tocopherol had no effect on rosemary's activity or a slightly negative one. Given that tocopherol turns pro-oxidant above its own optimum, a negative interaction is mechanistically plausible. Treat additive benefit as the reasonable expectation and synergy as an unproven bonus; the dose evidence for the tocopherol side is in vitamin E and antioxidants.
The chelator is the part that does not transfer. Trace iron and copper catalyse oxidation most efficiently in dry systems, so a balm is the worst case, but refineries add citric acid at around 50 ppm into hot oil during final processing and you cannot reproduce that by stirring powder into a cooled balm. Under-dosing a chelator is worse than omitting it: at a 1 to 1 molar ratio with iron, EDTA has been shown to promote oxidation rather than inhibit it. Exclusion beats chelation at bench scale, which means buying oils with low metal content and keeping the batch out of unlined metal.
- Deodorised rosemary extract at 100 to 150 ppm of carnosic acid plus carnosol.
- Mixed tocopherols at 0.05 percent, dosed on the active rather than the bottle strength.
- No chelator, because you cannot dose or dissolve one reliably in an anhydrous batch.
- Both added to the bulk oil when the bottle is opened, not to the finished melt.
- Oil phase kept below about 60 C after the antioxidant goes in.
Where it earns its place and where it does not
Rosemary extract is the most useful antioxidant available to a small maker, and it is still a marginal intervention. Every number on this page comes from bulk vegetable oils held at 110 or 120 C in a stream of air. There is no published induction time for a wax-structured anhydrous balm in a filled tin at 20 C that anyone should be quoting, because a wax network changes oxygen diffusion in ways the Rancimat does not model. What transfers is the direction and the rough magnitude, not a shelf life figure.
Three things beat it on return per unit of effort, and none of them is an additive. Choosing a less unsaturated oil removes the substrate: swapping a linolenic-heavy oil for high oleic sunflower changes the picture more than any dose can. Filling the container properly removes the headspace oxygen. Keeping the stock cool and dark removes the energy. Only after those three does an antioxidant defend a margin worth defending, and the relative weight of each factor for your own formula is what the shelf life calculator is for.
The decision rule, then. If your oil blend is mostly oleic, your tins are filled and your stock turns over in months, rosemary extract is optional and 100 ppm is plenty. If you are committed to a linolenic-heavy blend, use a deodorised grade at the top of the band, refrigerate the bulk oil, shorten the best before, and prove the result with matched samples rather than assuming it, following the protocol in shelf life testing. And if the product contains water in any form, none of this page applies to the problem you actually have.
Frequently asked questions
How much rosemary extract should I use in a balm?
Between 0.02 and 0.5 percent of the finished formula, depending entirely on the strength of the extract you bought. Aim for 50 to 250 ppm of carnosic acid plus carnosol in the product. A 5 percent extract needs 0.2 to 0.5 percent, a 10 percent grade 0.1 to 0.25 percent, and a 25 percent concentrate only 0.02 to 0.1 percent.
Is rosemary extract a natural preservative?
No. It is an antioxidant, so it slows oxygen attacking the fatty acids in your oils. Carnosic acid does inhibit some Gram-positive bacteria in laboratory assays, but at concentrations far above cosmetic use levels and with no useful action against Gram-negatives, yeast or mould. It is not on the EU list of permitted cosmetic preservatives, and a product containing water dosed with it will fail challenge testing.
Is rosemary essential oil the same as rosemary extract?
No, and substituting one for the other wastes the dose. The essential oil is steam distilled and contains volatile terpenes such as 1,8-cineole and camphor. Carnosic acid is non-volatile and does not survive distillation, so the essential oil holds essentially none of the antioxidant activity. They have different INCI names: Leaf Oil against Leaf Extract.
Will rosemary extract make my balm smell of rosemary?
A crude oleoresin will, noticeably, at anything above about 0.2 percent in an unscented product, and on a lip it reads as bitter rather than herbal. Deodorised grades are close to neutral at in-use levels, and deodorised plus decolourised grades add neither smell nor colour. Pay the premium for pale or unscented ranges.
What does the percentage on a rosemary extract certificate mean?
It is the assay for carnosic acid plus carnosol, the two diterpenes that do the work. Typical commercial strengths are 5, 10 and 25 percent, with purified fractions up to about 60 percent. The rest of the bottle is carrier oil. Two products both labelled rosemary extract can differ fivefold in active content, so the assay is the only number worth comparing.
Can rosemary extract save an oil that has already gone rancid?
No. Once hydroperoxides have broken down into aldehydes and ketones, those are stable end products and a radical scavenger has nothing to act on. An antioxidant delays the onset of oxidation, which is why food law describes antioxidants in prospective verbs such as retard and prolong. Adding extract to an oil that smells of crayon dilutes the smell and nothing more.
Should I use rosemary extract or vitamin E?
Both, at low doses, if you use either. They act on the same chain reaction by the same hydrogen-donating mechanism, and rosemary extract usually shows the larger induction time gain because most seed oils already carry substantial tocopherol and very little added alpha-tocopherol is useful. Synergy between the two is claimed widely and supported inconsistently, so expect additive benefit rather than multiplication.
Sources and further reading
- European Commission, Commission Regulation (EU) No 231/2012 laying down specifications for food additives, entry for extracts of rosemary (E 392), Official Journal of the European Union.
- European Commission, Regulation (EC) No 1333/2008 on food additives, Annex II, permitted levels for E 392 expressed as the sum of carnosic acid and carnosol.
- EFSA Panel on Food Additives and Nutrient Sources, Scientific opinions on extracts of rosemary (E 392), including the derived acceptable daily intake, EFSA Journal.
- Song, Sui and Jiang, Comparison of rosemary extract with synthetic antioxidants in vegetable oils, Foods, 2023;12:2177.
- American Oil Chemists' Society, Official Method Cd 12b-92, Oil Stability Index, AOCS, Urbana.
- Cosmetic Ingredient Review, Safety assessment of Rosmarinus officinalis (rosemary)-derived ingredients as used in cosmetics, CIR Expert Panel, Washington DC.
- US Food and Drug Administration, 21 CFR 182.20 Essential oils, oleoresins and natural extractives, eCFR.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.