The Layers of Your Skin, and What Reaches Each One

The layer that decides what any skincare product can do is 10 to 30 micrometres thick. The dermis underneath it is 1 to 4 millimetres. Most pages about the layers of skin list the names in order and stop there, which leaves out the question readers actually have: what reaches each one?

This article answers that, and audits one number on the way. The claim that your skin renews itself every 28 days is credited by one skincare brand to a 1977 paper. That paper's abstract reports 45 days.

Everything here describes structure and routes. It does not recommend a product or a treatment, and where I could read only an abstract or a review's summary of a study, I say so.

The Numbers First

The skin layers are usually described as three: the epidermis on top, the dermis beneath it, and subcutaneous tissue below that, which is mostly fat (review of dermal absorption). The epidermis has two working parts, the stratum corneum on the surface and the living epidermis under it. A review of the skin barrier gives thicknesses for each:

The thicknesses come from the skin barrier review, which also notes that the stratum corneum and the living epidermis both lack a blood supply and depend on the vascularised dermis beneath them. A micrometre (μm) is a thousandth of a millimetre.

Those ranges are averages, and on a face the average hides the interesting part. A 2025 study used high-frequency ultrasound at 75 MHz to scan 38 sites on the faces of 45 women aged 22 to 73. It measured the skin down to the fat beneath it, so epidermis and dermis together:

The study found that location explained about half the variation in thickness. Age explained less than 1%. Your upper eyelid is about 30% as thick as the tip of your nose.

Three limits on those numbers. All 45 participants were women, so the authors say the figures cannot be generalised to men. They were patients at aesthetic medicine clinics, which the authors flag as a possible source of outliers from earlier procedures or fillers. And the data are from one side of each face in a retrospective analysis. The authors declare no conflicts of interest.

There is also a disagreement the authors report themselves. Earlier thickness studies used biopsies from cadavers, and one of them ran about 30% lower than the ultrasound figures on average. The authors suggest measurement bias or method limits, and point out that postmortem tissue dehydrates and distorts thickness. Two earlier cadaver studies also disagree with each other. Take the absolute numbers as a good map of which sites are thicker than which, not as exact distances.

Two observations of mine, which are reasoning and not findings from either source. First, the scanner's axial resolution was 21 μm, which sits inside the stratum corneum's 10 to 30 μm range, so a scan like this cannot isolate the barrier layer from the rest of the epidermis. Second, the general dermis range of 1 to 4 mm cannot describe every site, because the whole of the eyelid's skin, epidermis and dermis together, is 573.5 μm. The two figures come from different methods and different sites, and both can stand.

The Outermost Layer Is the Whole Barrier

The stratum corneum is the first layer every product meets and, for most molecules, the layer that decides the outcome. It is made of dead, flattened cells embedded in a lipid matrix. According to the barrier review, that matrix is the only continuous pathway through the layer, so anything applied has to cross the lipid-filled spaces between cells. The lipids are ceramides, cholesterol and free fatty acids, with no phospholipids at all.

DermNet describes the surface layer as stacks of flat cells that move outward over about 14 days, loosen at the top and fall off, and readily absorb and lose water (DermNet on epidermal structure). How those lipids are lost and rebuilt is covered in our piece on the skin barrier.

The dermal absorption review describes two routes into skin. The main one runs through intact epidermis, either between the cells along the lipid matrix or through the cells themselves. The other runs through the openings of hair follicles, sweat glands and sebaceous glands.

The openings sound like a shortcut and are not much of one. The review says appendages account for only about 0.1% of the skin's surface area, and calls them minor routes, though relatively more important for large and ionised molecules. DermNet's rough per-square-centimetre figures, which it presents as approximations, are about 100 sweat glands, 15 sebaceous glands and 5 hairs. Plenty of openings, very little area.

The same review sets out the 500 Dalton rule: compounds with a molecular weight above about 500 Da struggle to cross normal stratum corneum, while smaller ones permeate more readily through the epidermal routes. It adds that peptides and proteins are too large to penetrate on their own. The review presents this as a working rule, not a wall, and our serums article walks through what it does to a typical shelf of ingredients.

The Living Epidermis: Where Pigment Is Made and Passed On

Under the barrier is the living epidermis, 80 to 200 μm thick, with no blood vessels of its own. It is fed from the dermis below, which means a product that reaches it has come from above, through the layer just described.

This is where new cells are made. Keratinocytes are created in the basal layer at the bottom and gradually move toward the surface, flattening and differentiating as they go. DermNet counts 5 to 12 layers of cells in the squamous layer above the basal layer and 3 to 5 in the granular layer above that, with a thin extra layer, the stratum lucidum, most obvious on palms and soles.

Pigment is made here too. Melanocytes sit in the basal layer at a ratio of about one to every ten basal keratinocytes, and they make melanin in packets called melanosomes. Their branching processes pass those packets to neighbouring keratinocytes. DermNet notes that all races have a similar number of melanocytes, and that what varies is how much melanin they make.

How Fast the Layers Turn Over: The 28-Day Figure, Audited

The claim is that skin renews itself every 28 days. One skincare brand's explainer states that the process takes approximately 28 days in healthy, youthful skin, and attributes the figure to Bergstresser and Taylor, published in the British Journal of Dermatology in 1977. It is the kind of number that gets repeated as a fact about skin, so it is worth finding out what the cited paper reports.

Bergstresser and Taylor built a model from desquamation rates of the stratum corneum. Their abstract says the model gives a total epidermal turnover time of 45 days for human volar forearms. I could read only the abstract, because the full paper is behind a paywall, so I cannot rule out a 28-day intermediate figure somewhere inside the model. The abstract does not contain one.

Weinstein and colleagues approached it differently in 1984, using tritiated thymidine to label dividing cells in normal human epidermis. Their abstract reports a mean epidermal turnover time of 39 days for the entire tissue. Again, I read the abstract, not the full paper.

So the two measurements I could open are 39 and 45 days. They differ from each other by six days, they use different methods, and both are longer than 28.

DermNet's epidermis page gives a spread. In one paragraph it says keratinocytes are normally renewed every 15 to 30 days, and that renewal is faster if the epidermis is injured and in certain diseases, particularly psoriasis. Further down it says basal cells divide on a cycle of roughly 19 days and that daughter cells take 28 to 60 days to reach the surface and form stratum corneum, with the stratum corneum itself taking about 14 days to move outward.

Those are different ranges on the same page, and the page does not say whether they describe different parts of the journey. The number 28 is the lower end of one of them and falls inside the other. I could not trace it to a measured total in any source I opened.

The 28-day figure is untraceable as a measured whole-epidermis value. The two totals I could verify are 39 and 45 days, and reference ranges run from 15 to 60. What survives is the general claim that epidermal renewal takes weeks, not days, and that it speeds up with injury and some skin diseases. A single number for everyone, in any state of health, does not.

The Dermis: Where the Structure Lives

The dermis is the thick layer, 1 to 4 mm in the general range. It is built on an extracellular matrix of collagen, and fibroblasts, its main cell type, make that collagen. It also carries the blood vessels, nerves and sensory receptors, helps regulate temperature and hydration, and supports the epidermis above it.

Claims about firming and wrinkles are claims about this layer, because the collagen is here. Of all the layers of skin, it is the one cosmetic claims most often point at, and it sits beneath the barrier, the size rule and the 0.1% appendage route.

Light gets there when molecules mostly do not. UVA reaches the dermis and is linked to collagen breakdown, which our article on SPF and wrinkles covers in detail.

For molecules, the picture is bleak. Our peptides article reports a study in which 0.22% of an applied peptide penetrated the stratum corneum, 99.7% was removed from the surface, and none reached the receptor solution, set against another study that found 30% penetrated within two hours. Those two figures differ by a factor of about 136.

The practical exception is a puncture. The dermal absorption review describes microneedles as arrays making transient holes through the stratum corneum, with needle lengths reported from 25 μm up to 1,500 μm, and describes microneedling as a way to bypass the barrier and deliver actives to the superficial dermis. That is a different category of thing from a cream, and our microneedling article covers why depth decides whether a device is doing anything at all.

The subcutaneous layer is mostly fat. The dermal absorption review adds that it holds abundant capillaries, which is how an active that gets far enough is carried around the body. The same review frames the goal of cosmetic delivery as reaching specific skin layers without entering the circulation, which is the line the section on the law below turns on.

A Depth Map of Common Actives

This table puts the skin layers on one scale and says what the sources support for each.

• Small molecules such as retinol, niacinamide, ascorbic acid, azelaic acid and glycolic acid are all under 500 Da, according to our serums article. In principle they can enter the epidermis. The size rule does not tell you whether they reach the dermis in useful amounts, and I did not find a source that measures that for a named product.
• Collagen, elastin and most proteins in creams are far above 500 Da, per the same article, so they are the likeliest to stay on the surface.
• Peptides sit between the two. Some are engineered with a fatty acid chain to improve penetration, and the measured data disagree by that factor of about 136.
• Device-delivered actives are a separate case, because the barrier is punctured. How deep the needles go decides the outcome.

This map is placement by molecular size and route, not a measured depth for any product. Statements about named ingredients are carried from our existing articles and the reviews they cite.

What "Penetrates Deep Into Skin" Can Legally Mean

The legal difference between a cosmetic and a drug in the United States turns on intended use. FDA's page on the question defines cosmetics as products applied to the body for cleansing, beautifying, promoting attractiveness or altering appearance. It defines drugs as products intended to diagnose, cure, mitigate, treat or prevent disease, or to affect the structure or any function of the body (FDA, cosmetic, drug or both).

Intended use is judged from the claims on the label, in advertising, on the internet and in other promotional material. The page says certain claims can make a product a drug even if it is sold as a cosmetic, and it gives examples such as claims to increase or decrease melanin production or to regenerate cells. It also states that the law does not recognise a category called cosmeceuticals, which our medical-grade skincare article covers further.

What the page does not do is mention penetration, absorption or depth at all. There is no definition of "deep", and no threshold of depth at which a product changes category.

My reading of that, which is reasoning and not a regulatory finding: collagen lives in the dermis, so a claim that a product rebuilds it is a claim about changing structure, and on the page's own wording that is drug territory. The more literally a product claims to act deep in the skin, the closer it moves to a drug claim. A phrase like "penetrates deep" on a cosmetic is therefore either loose marketing with no legal content, or a claim that needs the evidence a drug would need. I have not looked at how FDA has enforced this or at any court's reading, and this is the position of one FDA page, not a clearance of any product.

Conclusion

The layers of skin are a stack with very different thicknesses and very different gatekeeping. The stratum corneum is 10 to 30 μm and its lipid matrix is the only continuous way through it. The living epidermis beneath it, 80 to 200 μm, has no blood supply and makes both new cells and pigment. The dermis, 1 to 4 mm, holds the collagen, the vessels and the nerves.

Thickness is mostly about where on the body you are. On the face, site explained about half the variation in an ultrasound study of 45 women, and age under 1%. Your upper eyelid skin is about 30% as thick as the tip of your nose.

What reaches each layer is limited by that barrier, a size rule of about 500 Da, and an appendage route covering about 0.1% of the surface. Light reaches the dermis, and needles do. For most creams the dermis is the layer they are least likely to affect.

And the 28-day figure does not hold up as a measured total. The two totals I could open are 39 and 45 days, and DermNet's own ranges run from 15 to 60.