TL;DR: Caffeine constricts blood vessels and reduces fluid in the surrounding tissue. That is a real mechanism with measurable effects, which is why caffeine serums genuinely flatten morning puffiness. They do almost nothing for dark circles because most dark circles are not vascular, they are pigmentary or structural. The marketing conflates the two and sells one product as a fix for both, which means almost everyone with dark circles ends up with a serum that targets the wrong thing.
A reader emailed me a photo. She had been using a popular caffeine eye serum for four months. The text said: “My puffiness in the morning is much better but the darkness is exactly the same. Am I doing something wrong?”
She was not doing anything wrong. The serum was doing what caffeine does. Her dark circles were not the kind that caffeine treats.
This is the diagnosis problem I want to write about, because every eye serum on the shelf is sold to address both puffiness and dark circles as if they were the same condition. They are not the same condition. They have different causes, different anatomies, and different treatments. The same molecule cannot fix both.
What caffeine actually does
Caffeine is a methylxanthine. Applied topically at concentrations between 1 and 5 percent it has two main mechanical effects: vasoconstriction (it narrows small blood vessels) and lipolytic activity (it stimulates the breakdown of stored fat through phosphodiesterase inhibition). The vasoconstriction is the relevant one for the under-eye area. The lipolytic effect is what cellulite creams have always claimed, with weak evidence.
The Herman review of caffeine in cosmetics summarised the data on under-eye application (Herman and Herman, Skin Pharmacol Physiol 2013, PMID: 22336827). When caffeine is applied to the periorbital area, it constricts the small venules under the lower lid, reducing the volume of pooled blood and the interstitial fluid that gives morning puffiness its characteristic bulge. The effect is real and measurable within fifteen to forty-five minutes of application. It lasts a few hours.
This is why caffeine serums work for puffiness in the morning. The bulge under your eye when you wake up is a combination of pooled venous blood and interstitial fluid that accumulated overnight when you were horizontal. The caffeine reduces both. The skin sits flatter against the bone. The puffiness goes down.
This is real. The reviews converge on it.
The problem is that this mechanism does almost nothing for dark circles in most people.
The four types of dark circle
Dermatology has converged on a classification system for periorbital darkening that distinguishes between mechanisms (Huang et al., Int J Dermatol 2014, PMID: 24320079; Sarkar et al., J Clin Aesthet Dermatol 2016, PMID: 26962392). The types are:
Pigmentary. Excess melanin in the epidermis or dermis under the eye. This is the most common type in people with darker Fitzpatrick skin types and is often genetic. The darkness is brown and the skin under the eye is the same color whether you stretch it, change the lighting, or apply caffeine. The melanin is built into the tissue.
Vascular. Visible venous blood and interstitial fluid showing through thin skin. This is the type that responds to caffeine. The darkness has a blue or purple undertone, gets worse with poor sleep or dehydration, and visibly changes when you press on the area. This is also the type that produces morning puffiness alongside the darkness.
Structural. Shadows cast by the tear trough, which is a depression between the orbital bone and the cheek that deepens with age as fat pads shift and bone reabsorbs. This type is darker in overhead light and lighter in front-on light because it is genuinely a shadow, not a color. Caffeine does nothing. Most topicals do nothing. The fix is filler or surgery if you care enough.
Mixed. Two or three of the above combined, which is the realistic state for most adults over 35.
If you do not know which type you have, you cannot pick a product that has any chance of working.
The simplest at-home test: stretch the skin under your eye gently downward. If the darkness fades while stretched, it is vascular or structural. If the darkness stays the same color while stretched, it is pigmentary. Look at yourself in overhead light versus front-on light. If overhead makes it dramatically darker, you have a structural component.
Now look at the eye serum you bought. The marketing copy almost certainly addressed all four types as one condition. The molecule inside addresses one of them.
What caffeine cannot do
Caffeine does not affect melanin production. It does not penetrate to the basal layer in concentrations relevant to tyrosinase inhibition. It does not fill volume loss. It does not address the structural depression of the tear trough.
This means if your dark circles are pigmentary (most darker skin tones, much of South and East Asian populations, many North Africans and Mediterraneans), caffeine will reduce your morning puffiness and the darkness will be visually unchanged.
If your dark circles are structural (most adults over 40 with prominent tear troughs), caffeine will reduce your morning puffiness and the shadow will be visually unchanged.
If your dark circles are vascular and you are sleep-deprived or dehydrated, caffeine will do something. The vessels will constrict, the pooled blood will reduce, and the darkness will visibly fade. This is the small minority of dark-circle presentations where caffeine is a fix.
The mistake the marketing makes, and the mistake I think most consumers make, is to assume the serum will work because the serum is sold as working. The serum works for one mechanism. Most people have a different mechanism.
What does work for the other types
For pigmentary dark circles the best evidence is for a topical that reaches the basal layer with a tyrosinase inhibitor or with vitamin K. Hydroquinone 2 percent in a periorbital formulation can work but the FDA pulled OTC hydroquinone in 2020 and dermatologists are cautious about it under the eye because the skin is thin and absorption is high. Azelaic acid at 15 to 20 percent is gentler and modestly effective over twelve to twenty-four weeks. Retinoids at low strength (retinaldehyde or low-dose retinol) help over months by stimulating turnover and reducing pigment-loaded corneocytes. None of these are quick fixes. The pigment was deposited over years and reduction takes months.
For structural dark circles the only intervention that meaningfully changes the appearance is hyaluronic acid filler in the tear trough, performed by someone competent (Sarkar et al., 2016 review again). Topicals do nothing for shadow. Some people get partial improvement from skin thickening with sustained retinoid use, which reduces the contrast slightly, but this is a small effect.
For vascular dark circles, caffeine is the first-line topical. Cold compress works. Sleep works. Reducing salt the night before works. Wearing sunscreen daily reduces the photoaged thinning that makes the vessels more visible. Vitamin K topicals have weak data but some patients respond.
For mixed, you stack: a retinoid for pigment and tissue thickness over months, a caffeine serum for morning puffiness, and a structural fix if the tear trough is prominent enough to bother you.
This is not a single-serum problem.
The fingertip pressure test
I do this on every reader who emails about dark circles before recommending anything. Stand in front of a mirror with a single overhead light. Look at the darkness under your eyes.
Now press very gently with one fingertip on the darker area for three seconds and release. Watch what happens in the next thirty seconds.
If the area is paler for ten to twenty seconds and then refills back to dark, you have a strong vascular component. The pressure emptied the venules and the refill is what you see returning to baseline. Caffeine will help you.
If the area looks identical immediately after pressing as before, you have pigmentary or structural darkness. The pressure did nothing because there is no pooled blood to displace. Caffeine will not help you.
If the area looks paler immediately and stays paler for the full thirty seconds, you have edema (fluid) more than blood pooling. Caffeine helps. Diuretics and reduced salt help more.
This is a thirty-second test that the entire marketing apparatus of the eye-serum industry would prefer you did not do.
What I do for my own under-eye
I have moderate pigmentary darkness inherited from my mother’s side, mild structural depression in the tear trough that has worsened in the last five years, and mild vascular darkness on days I have slept badly. The eye area is mixed in the way most adults are.
I use a retinaldehyde 0.1 percent under the eye three nights a week. This is the only product I have used long-term that has produced measurable change in the pigmentary component. The pigment is about 30 percent lighter than it was three years ago, which is the kind of slow improvement that does not photograph dramatically but is real.
I use a caffeine serum in the morning on days I have slept under six hours or have an event in the evening. It flattens the morning puffiness within twenty minutes. It does not change the underlying darkness, which I have accepted.
I do not have filler in my tear trough. The structural component still casts a shadow in overhead light. I have decided that is fine for now. If it bothers me more I will reconsider.
The total spend on the under-eye area in the last three years is about $90. I do not own a $200 eye cream and I have not seen one outperform the retinaldehyde plus caffeine combination in any peer-reviewed comparison.
What I would tell my past self
The first thing to do with dark circles is diagnose the type. Not buy a serum. Not buy three serums. Stretch the skin, change the lighting, do the fingertip test. The five minutes spent on the diagnosis will save you years of buying products that target the wrong mechanism.
If the diagnosis is pigmentary, set expectations to twelve to twenty-four weeks of a retinoid and a tyrosinase inhibitor before you see meaningful change. If the diagnosis is structural, accept that topicals will not fix it and decide whether you care enough for a filler appointment. If the diagnosis is vascular, caffeine and sleep and salt management will help and the change will be visible within weeks.
The mistake is buying a multi-action serum because the marketing said it does everything. The molecule inside does what it does. The marketing is not chemistry.
Frequently asked
Does a $200 eye cream work better than a $20 one? Almost never. The difference between an expensive eye cream and a cheap one is the packaging and the marketing. The actives are usually similar concentrations of caffeine, peptides, niacinamide, or low-dose retinol. The cost-per-result is far better on the cheap end.
Is the skin under the eye really different from the rest of the face? Yes, it is thinner (about 0.5 mm versus 2 mm on the cheek) and has fewer sebaceous glands. This is why it shows vasculature more easily and why it tends to dry out faster. The thinness is also why retinoids need lower strengths there to avoid irritation.
Can I use my regular retinoid under the eye? Cautiously. Start with twice a week and a buffered application (apply moisturiser first, then a small amount of retinoid). Many people tolerate this. Some do not and need a lower-strength dedicated product like retinaldehyde 0.05 to 0.1 percent.
Do cooling rollers work? For morning puffiness, yes. The cold causes the same vasoconstriction as caffeine through a different mechanism. The effect is short and the tool is reasonable as long as you do not expect it to do anything for pigment.
Is poor sleep the main cause of dark circles? It can make vascular darkness worse for a few days. It does not cause pigmentary or structural darkness. If your circles persist on weeks of good sleep, sleep is not the issue and you need a different diagnosis.
Related Elelaf tools
Sources
- Herman A, Herman AP. Caffeine’s mechanisms of action and its cosmetic use. Skin Pharmacol Physiol 2013. PMID: 22336827.
- Roh M, Han M, Kim D, Chung K. Sebum output as a factor contributing to the size of facial pores. Br J Dermatol 2006. PMID: 16634875.
- Sarkar R, Ranjan R, Garg S, Garg VK, Sonthalia S, Bansal S. Periorbital hyperpigmentation review. J Clin Aesthet Dermatol 2016. PMID: 26962392.
- Huang YL, Chang SL, Ma L, Lee MC, Hu S. Clinical analysis and classification of dark eye circle. Int J Dermatol 2014. PMID: 24320079.