Vitamin E or rosemary extract: which antioxidant is better for oils?

Alpha-tocopherol peaks near 100 ppm in oil and turns pro-oxidant above it. Rosemary extract worked best at 50 to 200 mg/kg. Which to add, by oil, with doses.

For a balm made with ordinary plant oils, add rosemary extract, not more vitamin E. The reason is a number most comparisons leave out: the oil already contains vitamin E. In soybean oil, alpha-tocopherol protected best at about 100 ppm and turned pro-oxidant above that, and a refined supermarket rapeseed oil in a 2023 study arrived with 155.2 mg/kg of alpha-tocopherol before anything was added. Topping up an oil that is already past its optimum buys little and can make it worse. Rosemary extract brings a different molecule, carnosic acid, that the oil does not contain, and in the same rapeseed study it worked best at 50 to 200 mg/kg over six months at room temperature while slowing the loss of the oil's own tocopherol. Vitamin E is the stronger choice in two cases: an oil stripped of its tocopherols, and a product with a water phase.

• Alpha-tocopherol optimum: about 100 ppm
• Gamma optimum: about 300 ppm
• Natural mix optimum: 340-660 ppm
• Rosemary optimum, room temperature: 50-200 mg/kg
• EU food limit, vegetable oils: 30 mg/kg actives
• Rancimat gain with rosemary: 1.55-2.20 times

The two antioxidants, measured side by side

The first page of results for this question is two soap making forum threads, a cosmetic chemistry blog, a trade news item and a supplier's question and answer page. They give doses without saying what the dose is measured in, and they disagree on which is stronger without saying in what oil. The table below assembles what seven open sources measured: two tocopherol dose studies, two head to head comparisons of rosemary constituents against alpha-tocopherol, two rosemary extract studies in edible oils, and the European Food Safety Authority's opinion on rosemary extract as a food additive. Each row names its source.

Two rows decide most balms: the best dose row and the row above it. Vitamin E has a ceiling and most plant oils start near it. Rosemary extract has no ceiling in these sources and starts from zero. The rest of this page works through what the studies measured and then gives a choice for each kind of balm. Each material on its own is covered on the vitamin E and antioxidants and rosemary extract ingredient pages.

Why both answers circulate

Search this question and you get two confident answers. Forum posts say rosemary extract is the stronger antioxidant and vitamin E is barely worth adding. Other posts say vitamin E is the proven one and rosemary is a fashionable extra. Both are repeating real results. The results disagree because they were measured in different systems, and the system is what decides.

Where vitamin E won

Hopia and colleagues, a group that included Edwin Frankel, put carnosic acid, carnosol and alpha-tocopherol through the same tests in several lipid systems. In bulk corn oil triglycerides, they report, "alpha-tocopherol exhibited the most antioxidant activity followed by carnosic acid and carnosol." In every emulsion they tested, "alpha-tocopherol consistently exhibited more antioxidant activity than carnosol and carnosic acid."

The companion study by Frankel and colleagues used a commercial rosemary extract in tocopherol-stripped corn oil. There the extract, carnosic acid, rosmarinic acid and alpha-tocopherol were all "significantly more active than carnosol". In the corresponding corn oil emulsions, "the rosemary compounds were less active than in bulk oil". The authors explain this by the polar rosemary compounds partitioning into the water, where they protect the oil less.

Where rosemary won

The same Hopia paper found the reverse in bulk methyl linoleate, a single purified fatty acid ester: "carnosic acid was a better antioxidant than carnosol", and "both were more active than alpha-tocopherol." So within one paper, from one laboratory, the ranking flipped between two bulk lipids. The authors' conclusion is the most useful sentence in this literature: "The type and polarity of the lipid system used as a model substrate significantly affect the antioxidant activity of carnosol and carnosic acid."

The studies that report rosemary extract beating everything else are mostly run in ordinary edible oils, which arrive with their own tocopherols. Song and colleagues added rosemary extract to soybean, rapeseed, cottonseed, rice bran and camellia oils supplied "without additional antioxidants", meaning without anything added by the producer, not without the oils' natural vitamin E. They measured the oils' own tocopherols in the same paper. Every added gram of rosemary extract in that design is a new antioxidant arriving in an oil that already holds a stock of the old one.

Which applies to a balm

A balm is a bulk lipid with no water phase, made from oils and butters that carry their natural tocopherols unless someone has removed them. That is the edible oil case, not the stripped oil case and not the emulsion case. So the rosemary results transfer to a balm and the emulsion results do not. The stripped corn oil result still matters, because it shows vitamin E is a strong antioxidant in its own right. The reason it disappoints in a balm is not weakness. It is that the oil is usually already carrying about as much as helps.

The number that decides it: how much vitamin E your oil already holds

Evans, Kodali and Addis measured how much of each tocopherol protects soybean oil best, at 40 and 60 C, temperatures far closer to a warm shelf than a Rancimat block. Their results, from the published abstract:

Above those levels, the authors write, "the individual tocopherols and the tocopherol mixture exhibited prooxidation behavior that was more pronounced with increasing temperature from 40 to 60°C." A jar of balm on a sunny windowsill or in a glovebox is the higher of those two temperatures, not the lower.

Now set that against what an ordinary oil contains before you open it. The refined rapeseed oil Guo and colleagues bought from a local supermarket held 155.2 mg/kg of alpha-tocopherol. That is already one and a half times the alpha optimum Evans measured, and refined rapeseed is not an oil anyone thinks of as rich in vitamin E. Jojoba carries about 0.05 percent tocopherols and cosmetic argan 667 mg/kg, figures set out with their sources in argan oil against jojoba oil; both are inside or above the 340 to 660 ppm window for a natural mixture. Where your other oils sit is in the oil comparison table, but the certificate of analysis for your lot outranks any table.

More vitamin E can make it worse

Martin-Rubio and colleagues took the question further. They added alpha-tocopherol to soybean oil at levels from 0.002 to 5 percent by weight, stored the samples under accelerated conditions and followed the oxidation by proton NMR. Their finding: "except in the lowest proportion, α-tocopherol not only exerts a prooxidant effect on soybean oil but also modifies its oxidation pathway". They also report that the highest concentrations brought forward the formation of some toxic compounds, including oxygenated aldehydes and epoxides derived from the linoleic chains.

The lowest proportion in that series, 0.002 percent, is 20 ppm. The 0.1 to 1 percent of vitamin E often repeated in craft instructions is 1,000 to 10,000 ppm. On the evidence of these two studies, that instruction does not protect the oil better. It sits in the range where the tocopherol itself drives oxidation. Why the excess tocopherol starts new chains rather than stopping them is set out on the vitamin E page, along with the separate question of tocopheryl acetate, which cannot act as an antioxidant in the jar at any dose.

Rosemary extract: dose the active, not the bottle

Rosemary extract is sold at very different strengths, so a percentage on its own says little. The two studies that give a clear dose response both used concentrated extracts, and the strength changes how their numbers read.

Song, Sui and Jiang, 2023. They added a commercial extract with "a very high carnosic acid content (70%)" at 400 mg/kg to five vegetable oils and ran a Rancimat at 120 C. That is about 280 mg/kg of carnosic acid. Induction periods rose as follows, against a blend of BHA and BHT at 200 mg/kg in the same run:

Read that as a gain of 1.55 to 2.20 times from roughly 280 mg/kg of carnosic acid, in oils that already carried their natural tocopherols. It is a ranking at 120 C, not a shelf life. The synthetic blend was at half the dose by weight, so the comparison with BHA and BHT is not like for like either, a point taken up in synthetic antioxidants.

Guo and colleagues, 2023. This is the more useful study for a maker, because it ran three tests on the same oil and they did not give the same answer. The extract was 85 percent carnosic acid, added to refined rapeseed oil at 50 to 700 mg/kg.

• Rancimat, 110 C, 10 L/h of air. The oil stability index kept rising with dose. At 50 mg/kg it was 13.63 hours, 3.77 hours above the untreated oil; at 700 mg/kg it reached 25.38 hours. At 100 mg/kg the extract matched 50 mg/kg of TBHQ, a strong synthetic antioxidant.
• Oven storage at 65 C. On peroxide value over 20 days, the optimum was about 150 mg/kg. More was not better.
• Room temperature, six months, bottle opened. Each 4.5 kg sample was kept in a 5 litre plastic bottle at room temperature for six months, the way an opened bottle is used. The optimum was 50 to 200 mg/kg.

That gap is the practical lesson. The hot test rewarded every extra milligram; the realistic tests put the best dose at 200 mg/kg or less. Dosing to win a Rancimat run is dosing past what the oil needs on a shelf. In carnosic acid terms, Guo's room temperature window is about 42 to 170 mg/kg, and the 65 C optimum about 128 mg/kg.

Using both: what "synergy" actually rests on

The usual advice is to add both because they are synergistic. The evidence here supports a narrower and more useful statement: rosemary extract protects the oil's own vitamin E.

Guo found that adding the extract "delayed the degradation of endogenous α-tocopherol in rapeseed oil", and that 50 mg/kg of extract held off the complete loss of the oil's alpha-tocopherol by 10 days at 65 C. Hopia found that in corn oil "carnosic acid increased the oxidative stability of alpha-tocopherol", while carnosol, its own oxidation product, decreased it. So carnosic acid spares tocopherol, and the order of events matters: a fresh extract high in carnosic acid helps, and an aged one that has turned to carnosol may not.

None of that says you should add vitamin E as well. The oil already supplies it. Adding rosemary extract to an oil with its natural tocopherols is already a two antioxidant system, and on the Evans numbers, the second one is usually at or past its optimum before you start. Add tocopherol on top only when the certificate shows the oil is low in it, or when you have bought an oil that has been stripped or heavily processed. The case where a refined oil is less stable than its fatty acids suggest is explained in refined against unrefined shea butter, where refining removed gamma and delta tocopherols entirely.

The EU food limit, read correctly

Most dose advice for rosemary extract quotes the food additive rules, and most of it quotes them wrongly, because the limits are not for the extract. EFSA's 2018 opinion sets out the maximum permitted levels from Annex II of Regulation (EC) No 1333/2008, and states: "All MPLs are expressed as the sum of carnosol and carnosic acid, and some are expressed on the fat basis of the food."

Two consequences. First, a range such as "15 to 400 mg/kg" is a range of carnosic acid plus carnosol across all food categories, and quoting it as a percentage of rosemary extract understates the extract you would need by ten to twenty times for a 5 to 10 percent grade. Second, the bulk oil limits are low. For a polyunsaturated vegetable oil, 30 mg/kg of actives is below the bottom of Guo's best storage window. These are maximum permitted levels, set alongside an acceptable daily intake of 0.3 mg per kg of body weight per day, not doses chosen for the best protection. So the food limit is a useful reference for a lip product, which is partly swallowed, but it is not the dose at which the extract works best.

For cosmetics the position is different. The same opinion records that rosemary extracts are permitted in cosmetic products "as an antimicrobial, refreshing and tonic" according to the Commission's CosIng database, and that "according to the Regulation (EC) No 1223/2009 on cosmetic products, there is no limit." A function listed in CosIng is not an authorisation as a preservative, and nothing in these sources shows the extract preserving a product with water in it. How to word either antioxidant on a label is in reading a balm label.

Which to use, by balm

Every row below follows from the measurements above. The question to ask first is what the oil already holds, and the second is whether the product contains water.

Two rows need a word. The tallow row rests on the site's own sourcing: the USDA record cited on the tallow against shea butter page gives beef tallow 2.7 mg of vitamin E per 100 g, which is 27 ppm, well under the alpha optimum. That is the one common balm fat where adding a little tocopherol might help, and none of the studies here tested it, so rosemary remains the safer first choice. The saturated base row follows from chemistry rather than a test: squalane and mineral oil have no double bonds, and fractionated coconut oil has an iodine value of 0.5 at most, as set out in fractionated coconut oil against jojoba. An antioxidant protects unsaturated chains, and these have almost none.

A worked batch: 100 g of lip balm

Take a lip balm that is 15 percent beeswax, 25 percent shea and 60 percent sunflower and castor, so the oxidisable oil and butter phase is 85 g. The aim is about 100 mg of carnosic acid per kg of that phase, the middle of Guo's room temperature window.

1. Find the active needed. 100 mg per kg times 0.085 kg is 8.5 mg of carnosic acid.
2. Convert to extract. With a 10 percent carnosic acid extract, 8.5 mg of active is 85 mg of extract, 0.085 g. With a 25 percent extract it is 34 mg.
3. Make it weighable. A scale reading to 0.01 g cannot dose 0.034 g well. Dilute 1 g of extract into 9 g of the sunflower you are using and weigh ten times the amount: 0.85 g of the dilution for a 10 percent extract. The batch calculator scales this to other sizes.
4. Leave the vitamin E out unless the sunflower's certificate shows low tocopherol. If you add it to make a vitamin E claim, accept that it is not protecting the oil at that level.
5. Add it to the oil early. Stir it into the bulk oil when you open the bottle, not into a pot of oil that has sat warm for an hour. Neither antioxidant reverses oxidation that has already happened, which is the first point in rancid balm.

Whether that dose is enough for your formula is not something these studies can tell you. The way to find out is a split batch, half dosed and half not, stored side by side, as described in shelf life testing, or a faster but rougher version under heat in accelerated ageing. The shelf life calculator gives a starting estimate from the oil blend, not a guarantee.

The oils matter more than the antioxidant

Both antioxidants slow a reaction; neither removes what it reacts with. Song's own table shows the limit. Rosemary extract nearly doubled the Rancimat time of camellia oil, from 1.54 to 3.39 hours, and still left it short of untreated rice bran oil at 3.83. The fatty acids of an oil set its starting point, and an antioxidant moves it by a factor of about two in the best results here.

Choosing oils lower in linoleic and linolenic acid, as set out in rancidity and oxidation and carrier oils compared, changes the starting point itself. A high oleic oil in place of a standard sunflower does more than either antioxidant at any dose. Filling the container so there is little air above the balm does more again. The antioxidant defends the margin those choices create.

What these numbers do not settle

No study here tested a balm. Every measurement is in edible oils, corn oil triglycerides, methyl linoleate or emulsions. Beeswax, shea and the other solids in a balm were not tested with either antioxidant, and their own minor components may shift the result.

The optima are oil specific. Evans measured soybean oil and Guo measured rapeseed oil. The dose windows on this page are the closest measured figures, not constants. Different extracts also behave differently: Guo notes that differences in carnosic acid content between extracts "may lead to different antioxidant effects".

The head to head studies are old and few. The two direct comparisons of rosemary constituents against alpha-tocopherol date from 1996 and were run in model systems. The result they support is that ranking depends on the system, not that either wins in general. No study found compared mixed tocopherols against rosemary extract at their own optima in the same oil.

Hours are not months. A Rancimat time at 110 or 120 C ranks oils and treatments under one set of conditions. It does not convert into a shelf life, and Guo's own results show the hot test and the room temperature test choosing different doses.

Reviewed and updated 15 October 2026. Spotted an error? Tell us and we will fix and log it.