Anhydrous formulation: what you gain and what you give up
No water means no emulsifier, no preservative and no pH, but a new set of failure modes. The formulator's view of water-free skincare, with numbers.
Removing water from a formula is not a small change. It deletes about half of the standard cosmetic chemistry toolkit, along with about half of the standard problems, and replaces them with a different set that most formulators meet for the first time when a batch of shea balm turns to sand a fortnight after pouring. This page is the formulator's view: what actually disappears, what takes its place, and where the honest limits of the format are.
An anhydrous formula contains no water phase, so it needs no emulsifier, no preservative and no pH adjustment, and it cannot separate or grow mould. In exchange it feels heavier, cannot carry water-soluble actives or humectants without a solubiliser, and fails instead by oxidation, crystal change and oil migration.
What "anhydrous" actually excludes
Anhydrous means no water phase. In practice that rules out more than the obvious. Aqua and distilled water, clearly, but also floral waters and hydrosols, aloe vera juice (which is over 99% water), witch hazel, glycerites, aqueous extracts, honey (roughly 17-20% water), and fresh or incompletely dried plant material. It also rules out any ingredient supplied as a solution in water, which catches out a lot of people buying actives.
What is allowed: liquid oils, butters, waxes, fatty alcohols, fatty acids, silicones, esters, hydrocarbon bases, dry powders, oil-soluble actives, essential oils, fragrance and oil-soluble antioxidants. That set is smaller than it looks once you start trying to do anything other than occlude and soften, which is the honest constraint at the heart of the format. The structural consequences are covered in what is a balm.
What you no longer need, and what now matters more
| No longer needed | Why it goes | Now matters more | Why it arrives |
|---|---|---|---|
| Emulsifier | Only one phase, nothing to hold together | Structurant choice | Wax type and level are now the only thing controlling texture |
| Preservative | Water activity too low for microbial growth | Antioxidant | Oxidation replaces microbiology as the spoilage route |
| pH adjustment | pH is meaningless without water | Peroxide value and oil freshness | Incoming oil quality sets the shelf life of the batch |
| Homogenisation | Nothing to disperse | Cooling rate and pour temperature | Crystal form is set by how the batch cools, not how it is mixed |
| Phase-inversion control | No phases to invert | Polymorphism | Fats recrystallise into different forms over weeks |
| Microbial challenge testing | Low-risk under ISO 29621 reasoning | Stability, cycling and oil-migration testing | Physical stability is what actually fails in the field |
| Water-soluble actives | Nothing to dissolve them in | Oil-soluble actives and lipid choice | The lipid blend has to do all the work by itself |
The row that surprises people is the last one. In an emulsion you can carry niacinamide, panthenol, glycerin, urea, amino acids and a preservative system in the water phase, and the oil phase is doing a supporting job. In an anhydrous product the lipid blend is the entire formula. There is nowhere to hide a weak choice of oil.
The water activity reasoning, properly stated
Microbes do not need water in the sense of total water content. They need free water, the fraction that is chemically available rather than bound up. The measure is water activity (Aw), which runs from 0 to 1, where 1 is pure water.
- Most bacteria need an Aw of roughly 0.9 or above to grow.
- Most moulds need roughly 0.8, and osmotolerant yeasts and xerophilic moulds get down to roughly 0.6-0.7.
- A genuinely anhydrous fat blend sits far below all of those, which is why it does not need preserving.
- Honey is around 0.55-0.65, which is why honey keeps but honey in a balm is a risk if any moisture is introduced.
This is a threshold argument, not a spectrum argument, which is why "a bit of water" is a genuinely dangerous middle ground. A product with 3% water added carelessly does not get 3% of a microbial problem. It gets local regions of high water activity at the droplet interfaces, which is precisely where growth starts. Either commit to anhydrous or commit to a preserved emulsion. The full case is in do balms need preservatives.
When an anhydrous product does need a preservative
The exceptions are real and worth listing explicitly, because "anhydrous means no preservative" is repeated far too flatly online.
- Water-containing botanicals. Aloe, hydrosols, honey, fresh herbs, un-dried infusion material. Any of these puts free water into the tub.
- Shower and bath use. A cleansing balm or a shower product is exposed to running water and stored in a humid place. Many commercial cleansing balms carry a preservative for exactly this reason.
- Wet-finger dipping. A wide-mouth jar used with damp hands, especially a shared one, accumulates water and skin flora at the surface.
- Products containing an emulsifier. If a cleansing balm has 10% emulsifier so it rinses off, it will also readily take up any water it meets.
- Hygroscopic additives. Glycerin, sodium lactate and similar humectants pull water out of the air over time.
A tube or a squeeze tin with a small aperture is the cheapest risk control available. If you are selling, visible mould in a balm is nearly always traceable to one of the five routes above rather than to a failure of the base formula.
How to get water-soluble things in anyway
Glycerin. It will not dissolve in oil. Stirred straight into a melt at 3% it forms droplets that sink, pool at the bottom of the tin and bead out on the surface as the balm settles, giving a product that feels sticky in patches. To carry it you need a solubiliser or a small amount of a water-in-oil emulsifier (polyglyceryl-3 diisostearate and similar are the usual choices), typically at 1 part emulsifier to 1-3 parts glycerin, incorporated with high shear while the batch is molten. That works, and it is no longer really an anhydrous product in the risk sense.
Sodium hyaluronate. There is no version of this that works. The molecule functions by hydrating into a gel that binds water. A dry powder suspended in oil never hydrates, so it sits there as insoluble particles. Its presence on an anhydrous ingredient list is a marketing decision.
Panthenol and urea. Both are supplied water-soluble. Panthenol has a viscous liquid form that can be dispersed with a solubiliser at low percentages. Urea is a crystalline solid that must be dissolved in water to do anything; anhydrous urea suspensions are made industrially but are not a small-batch technique.
Powders. Clays, oxides, starches and zinc oxide disperse fine in oil and are the one class of water-loving material that behaves. They are also the reason many anhydrous products feel dry rather than greasy. See vitamin E and antioxidants for the oil-soluble additives that are actually worth including.
If a customer asks for "hydrating" from an anhydrous product, the honest formulation answer is to increase occlusion and improve the emollient blend, and to advise applying to damp skin. Applying a balm within three minutes of towel-drying traps water that is already there, which is the only hydration route an anhydrous product has.
The four failure modes that replace spoilage
Oxidation. Unsaturated fatty acids react with oxygen, producing the crayon or old-nuts smell of rancidity along with a sticky, tacky feel and sometimes a colour shift. Rate tracks polyunsaturate content: grapeseed at 65-75% linoleic acid has a 6-9 month practical life, high-oleic sunflower at 75-85% oleic gets 1-2 years, and jojoba, being wax esters rather than triglycerides, keeps 5 years or more. Antioxidants slow this; they do not stop it. Full treatment in rancidity and oxidation.
Polymorphism. Fats crystallise in several forms with different melting points and densities, and they convert slowly toward the most stable one. Shea butter recrystallising this way produces the classic gritty texture; cocoa butter produces pale streaks called fat bloom. Both are cosmetic rather than safety problems, both are prevented by full melting followed by fast cooling, and both are the reason a batch that looked perfect on day one can be unsellable on day twenty. See grainy shea butter.
Migration and sweating. The wax network holds liquid oil by capillary action. Overload it, or subject it to temperature cycling, and oil is expelled to the surface as beads or a wet sheen. Formulas with high castor oil or high liquid-oil fractions are the usual culprits, and ozokerite binds oil better than beeswax if you need a fix. See balm sweating.
Contraction faults. Molten balm is less dense than set balm, so every pour shrinks. Pour too hot and the shrinkage concentrates as a central dip, a tunnel down the middle of a tube, or a cracked top. This is a cooling-rate problem, not an ingredient problem, and it is solved by pouring lip tubes at 65-72 C and tins at 60-68 C.
The sensory ceiling
There is a limit to how light an anhydrous product can feel, and it is worth being honest about it rather than chasing it. Nothing evaporates from a balm, so there is no cooling and no sense of the product "sinking in" beyond the rate at which the oils spread and penetrate the outer stratum corneum. You can get closer to a light feel by choosing fast-spreading, low-viscosity esters and dry oils (caprylic/capric triglyceride, squalane, isoamyl laurate), by keeping wax low, and by adding powders to cut the gloss. You cannot get to the feel of a 70% water lotion. Anyone claiming otherwise is comparing against a bad lotion.
What the format wins in return is durability. An anhydrous product survives being carried in a pocket, left in a workshop, shipped in summer and used with dirty hands in ways an emulsion does not, and it can be made in a kitchen with a scale, a water bath and a thermometer rather than a homogeniser. For the texture consequences of each lever, see balm texture science; for the structural arithmetic, the wax ratio calculator does the conversions.
How to test something that cannot go mouldy
Stability testing for an anhydrous product looks nothing like a challenge test. What you are watching for is physical and oxidative change, so the useful protocol is: one sample at ambient, one at 40 C, one cycled between roughly 4 C and 40 C daily, one in the actual sales packaging on a windowsill. Check at 2, 4, 8 and 12 weeks for smell, surface beading, grain when smeared on glass, colour, and hardness by thumb pressure. Keep a retained sample of every batch you sell. The cycled sample will tell you about sweating and grain long before the ambient one does. Shelf-life testing sets out a workable schedule for a small maker.
If you take one habit from this page, make it the retained sample and the dated peroxide-value-by-nose check. Most anhydrous products do not fail dramatically. They drift, and the only way to notice drift is to have last quarter's batch sitting next to this one.
Frequently asked questions
What does anhydrous mean in skincare?
It means the product contains no water phase at all. Oils, butters, waxes, silicones, powders and oil-soluble actives only. Anhydrous is a structural statement about the formula, not a claim about quality or naturalness, and it is the defining property of balms, salves, ointments, body oils and anhydrous scrubs.
Do anhydrous products really not need a preservative?
A genuinely water-free product has water activity well below the roughly 0.6 that osmotolerant moulds and yeasts need, and far below the 0.9 most bacteria require, so nothing can grow. It stops being true the moment water gets in: a wet fingertip, a shower shelf, aloe, honey, a hydrosol or plant material that was not fully dried.
Can you put glycerin in a balm?
Only with help. Glycerin is not soluble in oil, so stirring it into a balm gives you droplets that separate, bead out on the surface and can feel sticky. It needs a solubiliser or a small amount of an emulsifier such as polyglyceryl-3 diisostearate to stay dispersed, and once water-attracting material is present the preservative question comes back.
Can hyaluronic acid go in an oil-based balm?
No, not in a functional way. Sodium hyaluronate is a water-soluble polymer that works by binding water in a hydrated gel. Dispersed as a dry powder in oil it cannot hydrate, cannot form its film and does nothing measurable. It appears on some anhydrous ingredient lists as a label claim rather than as working chemistry.
What is water activity?
Water activity (Aw) is the amount of water in a material that is free and available, on a scale from 0 to 1, rather than the total water present. It matters because microbes need free water. Honey has plenty of water by weight but low water activity, which is why it resists spoilage; an anhydrous balm has essentially none.
What are the main failure modes of an anhydrous product?
Four: oxidation of the unsaturated oils (rancid smell, sticky feel), polymorphic change in the fats (grain in shea, bloom in cocoa butter), oil migration out of the wax network (sweating and beading), and physical faults from cooling (dips, tunnels, cracks). None of them are microbial and none are prevented by a preservative.
Does vitamin E preserve a balm?
No. Tocopherol is an antioxidant: it slows the oxidation of unsaturated oils, typically at 0.1-0.5% of the oil phase. It has no antimicrobial function, it will not save a batch that has already gone rancid, and above roughly 1% it can behave as a pro-oxidant as well as adding colour and odour.
Sources and further reading
- US Food and Drug Administration, Microbiological Safety and Cosmetics, FDA.
- European Commission, CosIng cosmetic ingredient database, European Commission.
- United States Pharmacopeia, General Chapter <1112> Application of Water Activity Determination to Nonsterile Pharmaceutical Products, USP-NF.
- International Organization for Standardization, ISO 29621: Cosmetics, microbiology, guidelines for the risk assessment and identification of microbiologically low-risk products, 2017.
- Sato K, Crystallization behaviour of fats and lipids, a review, Chemical Engineering Science, 2001.
Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.