Under-eye bags and dark circles: four different causes that look alike

The skin of the lower eyelid is the thinnest on the body. Underneath it sits a ring of muscle, then a thin membrane, then the fat that cushions the eyeball, then bone. That is an unusually shallow stack, which means almost anything happening in any of those layers becomes visible at the surface. A little extra melanin shows. A vein a millimetre down shows. A dip in the contour shows, because the light has to come from somewhere and a dip means shadow. And a bulge shows, because a convex surface throws a shadow underneath itself.

All four register in a photo as darkness in roughly the same place. That is why the internet treats them as one condition with one fix, and why so much advice about it fails: people apply a surface remedy to a structural shape, or attribute a lifelong anatomical groove to last night's sleep. The distinctions below are the practically useful ones.

One thing to hold onto from the start: most faces have more than one of these at once, and the mixture changes what you see. A shallow hollow with pigment over it looks far darker than either alone. That is normal, and it is also why a single explanation almost never fits.

The first cause is the simplest: there is more melanin in the skin under the eye than in the skin next to it. Dermatology calls this periorbital hyperpigmentation. It is colour in the tissue itself, not an optical effect of shape or light.

It tends to read brown or tan rather than blue or grey, and its edges follow the skin — a defined patch or a band, sometimes extending onto the upper lid or beyond the orbital rim onto the cheek. It is strongly familial in many people, appears early rather than at forty, and is more commonly noticed in deeper skin tones, where melanocytes respond readily to any provocation. Sun exposure drives it, because the periorbital skin gets plenty of ultraviolet and rarely gets sunscreen. So does inflammation: eczema, contact reactions to cosmetics, and months of rubbing an itchy eye can all leave pigment behind after the irritation itself has settled. That is post-inflammatory hyperpigmentation, and it is the one variety of dark circle with an obvious behavioural contributor.

The defining behaviour of pigment is that it does not care about light. Move the lamp, tilt your head, stand under a window instead of a ceiling downlight — pigment stays exactly where it is and stays the same colour. That single property separates it from half of everything else in this article.

The second cause is not colour in the skin at all — it is colour seen through it. Beneath the lower lid runs a dense network of small vessels, and the orbicularis oculi muscle sits close to the surface. Venous blood, low in oxygen, is dark and reads bluish-purple when viewed through a translucent layer, in the same way surface veins on a pale forearm look blue rather than the colour blood actually is.

This variety reads blue, purple or slate rather than brown. It is more conspicuous the thinner and fairer the skin, and it becomes more conspicuous with age as skin loses thickness. It also fluctuates in a way pigment never does: anything that dilates those vessels or slows drainage makes it darker for a while. Crying, a cold room, congestion, alcohol, a bad night, hard exercise, a viral illness — all of it modulates the same anatomy without changing it.

A useful mental model: pigment is ink on the paper, vasculature is ink on the sheet underneath showing through. That distinction matters enormously later, because the surface approaches people reach for act on the paper.

This is the one most often misread, and the most important to get right. The tear trough is a groove that runs from the inner corner of the eye downwards and outwards along the lower orbital rim. It exists because the skin there is tethered fairly tightly to the bone by ligamentous attachment, with very little fat over it, while the cheek fat sits just below it and the eyelid tissue just above. A tether between two fuller areas produces a crease. That is a normal anatomical feature, present in plenty of people in their twenties.

A groove is not dark. A groove is a surface that faces away from the light, and what you are seeing is a shadow cast into it. The distinction sounds pedantic and is completely practical: shadows obey the light. Under a ceiling downlight, with the light coming from above and slightly behind, the trough fills with shadow and the area looks bruised. Under soft light arriving from directly in front at eye height, the same groove nearly vanishes. Nothing about the face changed. The illumination did.

Two things deepen a tear trough over time. One is loss of volume — the fat pad of the midface sits lower and holds less, so the step between eyelid and cheek becomes more abrupt. The other is skeletal: the shape and projection of the orbital rim differ between people, and a rim that sits back gives less support to the tissue in front of it. That second component is bone, and it is fixed. It is also completely invisible in a photograph, which is a recurring theme here.

If your dark circles disappear when a friend photographs you by a window with no direct sun on it and reappear in the bathroom mirror, you have your answer. That is a shadow, and the thing casting it is contour, not colour.

The fourth cause is the only one that deserves the word "bag". The eyeball sits in a socket cushioned by fat, and that fat is held back by a thin membrane called the orbital septum. With age the septum weakens and the fat behind it pushes forward, producing a genuine convexity in the lower lid. This is orbital fat prolapse, often called fat herniation.

Its defining feature is volume. It is not a colour and not a hollow — it is a shape that stands proud of the surrounding surface, visible in a three-quarter or profile view where a shadow would simply cease to exist. It usually casts its own shadow beneath itself, and because it very often coexists with a tear trough directly below, the two together produce the classic pouch-and-groove look that reads as tiredness even in a well-rested person.

It behaves differently from everything above. It does not resolve with sleep, it does not change with the light, and it does not respond to anything applied to the surface, because the tissue involved is millimetres beneath and structural. It can be more prominent in some families well before middle age. Whether a particular bulge is herniated fat, retained fluid, or something else entirely is a clinical determination made in person — several unrelated conditions can produce lower-lid swelling, and some of them are medical rather than cosmetic.

Layered on top of those four structures is a set of things that come and go. They are worth separating out because they are the only part of this that is genuinely under your control and costs nothing, and because they are responsible for most of the day-to-day variation people notice.

• Sleep and body position. Lying flat for hours lets fluid settle in loose tissue, and periorbital tissue is about as loose as tissue gets. That is why puffiness is typically worst on waking and eases within a couple of hours upright. Short or poor sleep also affects skin blood flow and colour, which makes the vascular component more visible.
• Sodium and alcohol. Both shift fluid balance. A heavy, salty meal or a night of drinking loads the compartment that shows it first. Alcohol additionally dilates vessels, so it can darken the vascular component and swell the tissue at the same time.
• Allergy. Histamine release causes vasodilation and tissue swelling, and nasal congestion impedes venous drainage from the region — the combination is well known enough to have a nickname, allergic shiners. Seasonal darkening that arrives with pollen and leaves with it is a strong hint.
• Rubbing. This is the sneaky one. Rubbing an itchy eye feels reversible, but repeated friction on the thinnest skin on the body is exactly the provocation that leaves pigment behind. A temporary cause quietly converts itself into a permanent one.
• Crying, illness, dehydration, hormonal fluid shifts and heat. All temporary, all capable of making a photo taken on the wrong afternoon look like a diagnosis.

The reason to catalogue these is not to promise that fixing them fixes everything. It is that they give you a timescale test. Anything that changes across a day or a week is a reversible contributor. Anything present in every photograph of you for the last five years, at every hour, in every light, is structure. Those two categories deserve completely different responses, and a lot of frustration comes from treating one as the other.

A large share of the under-eye darkness people worry about is manufactured at the moment of capture. We are not going to put a number on how large, because nobody honestly can. But this is not a consoling platitude; it is a statement about how shadows work, and it is testable in about ninety seconds.

Light direction. Overhead light — ceiling downlights, midday sun, most office and bathroom fittings — arrives from above the brow. The brow ridge blocks it, the eye socket is recessed, and the entire orbital region drops into shadow. This is the single harshest configuration for the under-eye area that exists, and it is also the one you meet most often, which is why the mirror in a badly lit bathroom is such a reliable source of despair. Move the same face to soft frontal light at eye height and much of it fills in.

Head pitch. A shadow depends on the angle between the light and the surface, so tilting your chin down or up changes the shadow without changing anything else. Tipping the chin down under an overhead source deepens the whole region; lifting it flattens it. Every self-taken photo bakes one arbitrary head angle in permanently.

Camera height and distance. A phone held above eye level and close in — the default selfie geometry — combines an unflattering shadow angle with noticeable perspective distortion of the features nearest the lens. Neither has anything to do with your tissue.

Processing. Modern phones do not record a scene, they reconstruct one. Local contrast enhancement and sharpening act hardest on soft-edged, mid-tone transitions, which is a precise description of an under-eye shadow. The same face, same light, two phones, two different degrees of darkness. This is the same problem we describe at length in why before-and-after photos of teeth are unreliable, and it applies just as forcefully here.

A mirror and a photograph also differ in a way people rarely account for: a mirror shows you under the room's current light and you unconsciously adjust your head until you look acceptable. A photograph freezes one instant of one lighting setup and one head angle, chosen by whoever pressed the button.

These are observations, not tests in a clinical sense. They will not tell you what you have. They will often tell you what you are mostly looking at, which is enough to stop you wasting effort on the wrong category.

1. The relight test. Stand under an overhead light and look in a mirror. Then hold a lamp or a phone torch at eye height, directly in front of your face, and look again. Darkness that largely disappears was a shadow. Darkness that survives is in the skin or under it.
2. The stretch test. Gently pull the skin at the outer corner sideways so the area flattens, without pressing on the eye. Pigment stays with the skin and remains visible. A contour shadow flattens out because you have removed the groove.
3. The blanch observation. Vascular darkness momentarily pales when light pressure empties the vessels and returns as they refill. Pigment does not change. Be gentle; the eye is not a place to press hard.
4. The look-up test. Keeping your head still, raise your eyes toward the ceiling. This puts the lower lid on stretch and often makes a genuine fat bulge more obvious against the changed contour.
5. The morning-versus-evening test. Photograph yourself under the same lighting on waking and again eight hours later. Fluid-driven puffiness settles noticeably. A structural bulge is there in both.
6. The oblique view. Look at yourself in a three-quarter view. Volume shows as a silhouette change. A shadow cannot exist in a silhouette, so if it is still visible in profile, it has substance.
7. The colour read. Brown suggests pigment; blue or purple suggests vasculature; neutral grey that tracks the light suggests shadow. Weak evidence on its own, useful alongside the rest.

Run all seven and you will usually end up with a sentence like "mostly a shadow, with some pigment, and it gets worse in spring". That sentence is a far better starting point for a conversation with a professional than "I have bags".

It is worth being exact about our own limits, because this is a topic where a measurement tool could easily overclaim. RealSmile places 68 landmarks on a photograph and derives 17 facial-geometry metrics from them, then expresses the result as a Face Score percentile. Landmarks mark positions. They record where the corner of an eye is, not what the tissue beneath it is made of.

The metric closest to this subject is Brow-Eye Proximity, which is the vertical gap between the brow and the eye opening expressed as a ratio scaled to face size. It describes the geometry of the upper eye region — how open or compressed the brow-to-lid space reads. It says nothing whatsoever about the lower lid. It cannot see melanin, it cannot see a vein, it cannot see a fat pad, and it cannot see the orbital rim behind all of it. Neighbouring metrics like Facial Thirds, Midface Ratio and FWHR are the same kind of thing: proportion, not tissue.

Anywhere we describe skin, we are describing photo appearance and nothing more — the pixels in the image you uploaded, under the light you happened to be standing in, processed by the phone you happened to be holding. Given everything in the section above about lighting, that is exactly the quantity this article has spent several hundred words warning you not to over-read. We report it as appearance because that is all it is.

There is a related honesty point from our own data that we publish rather than hide: a Face Score is photo-dominated. The same face, rescanned minutes later under different light and at a different camera height, moves by several points. If a purely geometric reading is that sensitive to capture conditions, an appearance judgement about a soft shadow under an eye is far more so. On /analyze we read one thing about the eye region with reasonable confidence — whether the orbicularis oculi is activated in a smile, the Duchenne marker that distinguishes a felt smile from a performed one. That is a read on an expression, not on tissue.

The whole clinical task in this area is the same separation this article describes, done properly and in person. A doctor examining a lower lid is not looking at a photograph. They are looking at a live face under directional light they control, from several angles, upright and often reclined, with the option to touch the tissue and see how it moves.

The assessment generally covers where the darkness sits in the tissue depth, how much of it is contour versus colour, the quality and thickness of the skin, the position and support of the lower lid itself, the projection of the underlying bone, and how the midface volume sits relative to the orbital rim. Alongside that sits history, which a photo can never supply: how long it has been there, whether it came on suddenly, whether it is symmetrical, whether it runs in the family, whether there is allergy or eczema, what medications are involved, how sleep and general health are.

Some findings shift the conversation out of cosmetics entirely. Swelling that appears abruptly, affects one side only, comes with pain, redness or fever, changes vision or the position of the eyelid, or accompanies swelling elsewhere in the body, is a reason to see a doctor promptly rather than to research it further. Persistent puffiness can have systemic causes, and distinguishing those from cosmetic ones is precisely the judgement we are not qualified to make and you should not attempt from a mirror.

We are not going to tell you to do anything. What is useful, and what almost no one lays out, is how the categories of approach map onto the categories of cause — because a mismatch guarantees disappointment regardless of how good the approach itself is.

Reversible contributors respond to reversible changes. Sleep, sodium, alcohol and managing an allergy with a clinician address the fluctuating layer. They are the cheapest part of this list and the part with the least to lose — though allergy management is itself a clinical matter, and anything you might take for it belongs in a conversation with a doctor rather than in an article. They will not flatten a herniated fat pad or remove lifelong pigment, and anyone claiming otherwise is selling something.

Surface approaches act on the surface. Anything topical acts on the skin layer, so it is only relevant where the problem is in the skin — pigment, texture, thickness. Applied to a contour shadow or a fat prolapse, it is acting on the wrong tissue entirely. Whether any particular topical approach suits your skin is a question for a dermatologist; the periorbital area is thin, easily irritated, and close to the eye, which is exactly why it is not a place for experimentation. The habit named most consistently in dermatology for pigmentation of any kind is sun protection, because ultraviolet exposure drives pigment broadly and the skin around the eyes is routinely left unprotected — what is suitable this close to the eye is, again, a question for a clinician.

Volume approaches address a hollow, not a colour. Filling a groove changes the contour so it stops catching shadow. It does nothing to pigment and nothing to a bulge above it. The under-eye is widely regarded as one of the more technically demanding areas in aesthetic practice precisely because the skin is so thin that irregularities show, and complications in a region this close to the eye are taken seriously. Some materials are described as reversible, but reversal is itself a procedure with its own considerations, and "reversible" is not the same as "undoable without consequence". This is operator-dependent work and the assessment of whether it is appropriate at all belongs with a qualified specialist.

Surgical approaches remove or reposition tissue, and removal is irreversible. Lower-lid surgery addresses herniated fat and, sometimes, excess skin. Once tissue is removed it does not come back. Over-removal is a well-recognised failure mode that can leave the area looking more hollow and more aged than before, and lid position and shape can be affected. There is also the question of what happens over the following decades, since the face keeps changing after any operation. Everything about whether, when, and how much is a specialist judgement made on an individual anatomy — not something to be decided from an article, a photo, or a scan.

Energy and pigment-directed procedures carry a pigment risk of their own. Any procedure that provokes inflammation in skin prone to hyperpigmentation can leave more pigment than it removed. This risk is not uniform across skin tones, and it is one of the most important things to raise directly with the person assessing you.

The through-line: treating a shadow as though it were pigment does nothing. Treating pigment as though it were volume does nothing. Treating fluid as though it were structure works for a morning and then stops. Identifying the dominant cause is the entire exercise, and it is the part most people skip.

A consultation goes better when you arrive with the right questions rather than a request for a specific intervention. These tend to produce clear, informative answers, and the quality of the answers tells you something about the person giving them.

• Which of these components is dominant in my case — pigment, vasculature, contour, or volume — and how did you determine that?
• How much of what I am seeing is shadow that would change under different lighting?
• Is there anything here that suggests a medical rather than a cosmetic cause?
• What happens if I do nothing at all?
• Which of the options you have mentioned are reversible, and which are permanent?
• What does a poor result from this look like, how often does it happen, and can it be corrected?
• What is the specific pigment risk for my skin type?
• What would you expect to improve with no procedure at all — sleep, allergy management, sun protection — and how long should I give that first?
• How will we record a baseline, so that any change is measured rather than remembered?

That final question is the one people forget. Memory for faces is unreliable and memory for a subtle change in a soft shadow is worse. Without a controlled baseline image, in three months the only comparison available is between a new photo and a recollection, and the recollection will agree with whatever you are hoping.

If you are going to track this over time, or show it to someone, the photograph needs to be a measurement rather than a mood. The principle is the same one every measurement discipline uses: fix the instrument, fix the conditions, and keep a reference.

1. One soft light, from the front, at eye height. A window with no direct sun is ideal. Switch other sources off — mixed lighting makes colour unstable and shadows doubled.
2. Camera at eye level. Not above, not below. Prop the phone rather than holding it out, and use the rear camera, which is the better sensor.
3. Lock exposure and white balance. Auto everything defeats the purpose, because the phone will make a fresh decision each time and you will read that decision as a change in your face.
4. No portrait mode, no beauty filter, no retouching. Some of these actively smooth the exact region you are trying to observe.
5. Same time of day. Morning and evening are two different faces where fluid is concerned. Pick one and stay with it.
6. Take four frames. Straight on with a neutral expression; straight on looking up; a three-quarter view; and one under deliberately overhead light for comparison. The set tells you far more than any single image.
7. Keep the originals and compare side by side. On one screen, at one brightness, in one session — never against memory.

The same discipline improves anything you measure from a photo, which is why our tools push you toward even, frontal light and a level camera before they push you toward anything else. If you want to see how much capture conditions move a purely geometric reading, run the 17-metric scan twice under different lighting and compare the numbers. It is an instructive experiment and it makes the point better than we can in prose.

The under-eye is not the only part of the face where one appearance has several unrelated causes. The pattern repeats, and recognising it saves a great deal of wasted effort.

A soft submental area is the closest parallel: it can be subcutaneous fat, it can be lax skin, it can be the position of the hyoid and the length of the jaw behind it, or it can be head posture in the moment the photo was taken. Those four look nearly identical from the front and are as different from each other as pigment is from fat herniation. We unpack that one separately in what a double chin actually is, and the reasoning is deliberately the same shape as this article.

Jaw definition works the same way. What reads as a sharp or soft jawline in a photo is some combination of underlying angle, soft tissue over it, head pitch and the direction of the light — which is why our jawline measurement reports an angle from landmarks and stops there, rather than pretending to know which of those components produced it. Same epistemics, different region.

The general lesson is that appearance is an output of several independent inputs, and a photograph collapses them into one image. Any tool, article or advertisement that reverses that collapse confidently is guessing.

Under-eye darkness is close to universal. It is a predictable consequence of having the thinnest skin on the body sitting over vessels, in a recess, under a brow that blocks overhead light. Almost everyone has some version of it, and a version that looks alarming in a still photograph is frequently not noticed at all in ordinary conversation, because there the light is moving, the head is moving, and nobody is examining a frozen frame at twenty centimetres.

There is also a straightforward observation about what actually moves a first impression, and it is not the tone of the skin beneath the lower lid. It is whether the eyes engage — whether a smile reaches them or stops at the mouth. That is a muscular event, it is free, it is reversible in both directions, and it is the one thing in this entire region we will make a confident claim about. If you want that read on your own face, it is what /analyze does, and the Face Score percentile gives you the geometric context around it.

If you take one thing away: name the cause before you chase the fix. Relight the area and see what survives. Track it across a day. Photograph it properly. Then, if it still matters to you, take that far better description to someone qualified to examine it in person. That sequence is cheap, it is honest, and it protects you from spending money on the wrong category of solution.