Double cleansing: where the evidence actually lands

Double cleansing has become a step people feel faintly guilty about skipping. Oil or balm first, foaming or gel cleanser second, every night. It gets presented as basic hygiene, so we went looking for the research behind it.

Nobody has trialled the routine itself

We could not find published research on double cleansing. No trial of an oil cleanser followed by a foaming one, in either direction.

That doesn't make the routine wrong. Plenty of sensible habits have never been trialled. But when a brand quotes a percentage of pollution removed or sunscreen left behind, that number isn't coming from a study of double cleansing. What follows is either chemistry or evidence from adjacent questions, and we'll say which is which.

Different grime comes off in different ways

The most useful paper we found tested five surfactants against soils on an artificial skin model, and two physical mechanisms showed up (Slavova et al., 2025). Sebum came off by emulsification, and a sodium laureth sulphate solution did that well. Dimethicone, a silicone, came off only by roll-up, a mechanism driven by how strongly the surfactant sticks to the surface. Coco glucoside, a nonionic surfactant, shifted dimethicone at a lower concentration than the ionic ones did.

Here is the core of the two-step argument. Silicones are common in long-wear foundation and water-resistant sunscreen, and they don't behave like sebum in the basin. One surfactant system doesn't lift every soil equally well. The limits are real though: artificial skin in a laboratory, soils applied for the test, and no comparison of an oil cleanse followed by a foam.

Sunscreen and long-wear makeup

Marketing leans hardest here, and what sits underneath it is thin. Liao et al. (2020) built an analytical method for measuring UV filters and, in the process, quantified five of them, including avobenzone and octocrylene, in human stratum corneum after simulated everyday sunscreen use. The authors concluded that a makeup remover should be used to take water-in-oil sunscreens off the skin.

That is the closest thing to direct support for an oil phase, and the chemistry is coherent. It is also a method paper: no comparison group, no measurement of one cleanse against two, and a recommendation rather than a test.

About those pollution percentages

The study behind pollution removal claims used an artificial soil of atmospheric particulate mixed into human sebum, applied to the cheeks of 20 people, cleansed for five seconds, then scored by photographic pixel intensity (Peterson et al., 2017). A sonic brush removed significantly more particulate than washing by hand (P less than 0.001). A device against bare hands, on a laboratory soil, no skin outcome measured, no oil-then-foam sequence in the design. A percentage built on that tells you very little about skin.

The surfactant is what has actually been measured

Here there is real measurement, though still at small scale. Surfactants cost the skin barrier in two ways. They bind and denature proteins in the stratum corneum, which is that tight feeling after washing, and they pull out intercellular lipids. Surfactants that interact less with skin protein are milder, and cleansers near skin's own slightly acidic pH are potentially less damaging than high-pH ones (Ananthapadmanabhan et al., 2004).

A recent review sorts surfactants by class (Abdollahi et al., 2025). Charge and micelle size drive irritancy. Anionic surfactants sit at the irritating end of the ranking, with amphoteric and nonionic types further down, and amino-acid-based and polymeric ones lower again. The review flags newborn and atopic skin as more vulnerable.

Twenty-nine cleansers applied to the forearms of 30 people over five days produced a significant correlation between cleanser pH and the irritation index (P less than 0.006), and the least irritating were mostly syndets, the synthetic detergents (Baranda et al., 2002). Thirty people, five days, forearms: that is the scale of the human evidence.

You often can't check what you are buying, which is the frustrating part. A survey of 250 marketed cleansers found all 37 soaps were highly alkaline, 84.9 per cent of liquid syndets were acidic, and only 12.8 per cent disclosed their pH at all (Khan et al., 2024). Format tells you more than the word gentle on a bottle: a liquid syndet is more likely to sit near skin pH than a bar soap.

Something awkward falls out of that. Two washes with well-formulated low-pH syndets might cost the barrier less than one wash with an alkaline soap. Nobody has tested it, so treat it as reasoning, but the count of washes looks like the wrong argument.

Two trials on washing, in different skin

One caveat travels with both: they studied washing in eczema-prone skin, not facial double cleansing, so they apply by analogy only.

In the WASH study, 43 children with atopic dermatitis in remission washed one limb with water only and the other with their usual cleanser, alkaline soap or acidic detergent, over roughly eight weeks in summer. Median eczema severity scores were 0.00 for water and 0.15 for cleanser (P equals 0.74), and the trial did not detect a difference in symptom scores, in skin infections, or in how often the children needed rescue treatment, which is prescription medicine a doctor manages and nothing a retailer or salon supplies (Katoh et al., 2024). It was a non-inferiority design in 43 children, not a demonstration that the two are equivalent.

A pragmatic online trial then randomised 438 people with eczema to weekly or daily bathing for four weeks and found no evidence of a difference in symptoms (P equals 0.30) (Bradshaw et al., 2026). Four weeks is short, bathing is whole body, and a null result at that size means a difference was not detected, not that none exists.

In those two trials, in eczema-prone skin, the extra washing produced nothing measurable. Neither added a second facial cleanse, so what they dent is the broad assumption that more washing buys you something, not this routine in particular.

Where cleansing stops

Nothing in this literature supports cleansing as a treatment. The only randomised trial of washing frequency in acne we found had people wash once, twice or four times daily with a mild cleanser for six weeks (Choi et al., 2006). Between the groups it found no significant difference. Within groups, twice-daily washing showed improvement in open comedones and non-inflammatory lesions, while once-daily washing showed increases in redness, papules and inflammatory lesions. Those are movements inside a small, male-only, six-week trial. A systematic review of face washing in acne concluded the evidence base was very limited and that advice should be individualised by a clinician (Magin et al., 2005).

Acne is a medical condition and skincare is not how it gets managed. If skin is persistently sore, painful, scarring or simply not settling, that is a conversation for a doctor, not something a cleanser answers.

On harm, there is no published study of damage from double cleansing, so anything said here is inference from the surfactant work above. No trial has measured barrier harm from repeated facial cleansing, and the only frequency trial we found was small, short and run in acne, so we cannot tell you what the number of steps costs.

Where we have landed

The chemistry case for an oil or balm step is real and indirect: silicone behaved differently from sebum in the one artificial skin study available, and the sunscreen evidence is an author recommendation rather than a comparison. The nightly oil-then-foam ritual for everyone has never been tested, and the numbers used to sell it come from studies of something else. For anyone wearing long-wear makeup or heavy water-resistant sunscreen, a first step built to dissolve those has coherent chemistry behind it. For skin that runs dry and easily irritated, nothing here puts one mild low-pH wash behind. The evidence base is thin either way.

References

1. Slavova et al. (2025). Cleansing Mechanisms and Efficacy on Artificial Skin. Molecules (Basel, Switzerland). PMID: 40333845
2. Liao et al. (2020). Ultrasound-Vortex-Assisted Dispersive Liquid-Liquid Microextraction Combined with High Performance Liquid Chromatography-Diode Array Detection for Determining UV Filters in Cosmetics and the Human Stratum Corneum. Molecules (Basel, Switzerland). PMID: 33053784
3. Peterson et al. (2017). A robust sebum, oil, and particulate pollution model for assessing cleansing efficacy of human skin. International Journal of Cosmetic Science. PMID: 27797421
4. Ananthapadmanabhan et al. (2004). Cleansing without compromise: the impact of cleansers on the skin barrier and the technology of mild cleansing. Dermatologic Therapy. PMID: 14728695
5. Abdollahi et al. (2025). Surfactant-containing detergents: Impacts on dermal health. Colloids and Surfaces B: Biointerfaces. PMID: 40795538
6. Baranda et al. (2002). Correlation between pH and irritant effect of cleansers marketed for dry skin. International Journal of Dermatology. PMID: 12207765
7. Khan et al. (2024). Not all marketed skin cleansers' pH is optimal for atopic dermatitis. Allergy and Asthma Proceedings. PMID: 38982600
8. Katoh et al. (2024). Skin care by washing with water is not inferior to washing with a cleanser in children with atopic dermatitis in remission in summer: WASH study. Allergology International. PMID: 38310043
9. Bradshaw et al. (2026). Weekly versus daily bathing for people with eczema: results of the Eczema Bathing online randomized controlled trial. The British Journal of Dermatology. PMID: 41208015
10. Choi et al. (2006). A single-blinded, randomized, controlled clinical trial evaluating the effect of face washing on acne vulgaris. Pediatric Dermatology. PMID: 17014635
11. Magin et al. (2005). A systematic review of the evidence for 'myths and misconceptions' in acne management: diet, face-washing and sunlight. Family Practice. PMID: 15644386