Skincare ingredients for dark circles
Our research data links 24 ingredients to dark circles. Strongest evidence comes first, each with the studies behind it.
A research list, not medical advice. If a skin condition worries you, see a dermatologist.
Which ingredients have the most evidence for dark circles?
In our data, the strongest evidence for dark circles is behind Sodium ascorbate, L-Ascorbic Acid, and 4-(1-Phenylethyl)resorcinol.
Get a free dark circles routine matched to your skin
Answer 5 questions, skip the rest. Free, and no sign-up to see it.
| Ingredient | Evidence | Studied at | Rating |
|---|---|---|---|
| Sodium ascorbate | 🧪 Clinical evidence | 10% | Low concern |
| L-Ascorbic Acid | 🧪 Clinical evidence | 20% (uncontrolled), 5% (other use) | Low concern |
| 4-(1-Phenylethyl)resorcinol | 📄 Literature evidence | Study dose: not found yet | Low concern |
| Caffeine | 📄 Literature evidence | Studied only mixed with other actives | Low concern |
| Niacinamide | 🧪 Clinical evidence | 2-5% (other use) | Low concern |
- Sodium ascorbateLow concern
Studied at 10%
Ohshima 2009 split-face vehicle-controlled (n=14, 6 months): the 10% sodium ascorbate side had a smaller change in erythema index (visible vessels) than vehicle and a tendency to thicker dermis - the authors interpret this as thickening the dermis to mask congested color. Because of the small sample it cannot be given above 4.5. The only vehicle-controlled single-ingredient evidence for dark circles.
- Reference: Ohshima et al., 2009
Pregnancy: – · Listed in 70 products
- L-Ascorbic AcidLow concern
Ascorbic Acid
Studied at 20% (uncontrolled), 5% (other use)
One trial measures this directly for pigment-type under-eye circles. Dayal 2016 (90 people in 3 groups, 12 weeks): in the group that applied 20% vitamin C daily, 26.7% improved by more than half, with the fewest side effects. The in-office peels in the same trial (glycolic 20%, lactic 15%) worked better. There is no untreated group, so part of the change may be placebo or sun avoidance, and the abstract does not give the chemical form of the vitamin C. For vascular dark circles there is no direct evidence. Sodium ascorbate has firmer single-ingredient evidence because its trial had a vehicle group. The form of the vitamin C is not confirmed, so the result cannot be read as the effect of one form. The full paper is behind a paywall and was not read.
- Reference: Humbert et al., 2003
- Reference: Espinal-Perez et al., 2004
- Reference: Murray et al., 2008
- Reference: Dayal et al., 2016
Pregnancy: – · Listed in 2,002 products
- 4-(1-Phenylethyl)resorcinolLow concern
Phenylethyl Resorcinol
Study dose: not found yet
Yu 2021: RCT, 20 women, a gel without the active as the control; PR gel once daily on one under-eye side and PR gel plus nano-microneedling twice a week on the other. Melanin index fell below baseline by week 8 in the PR group (more on the microneedle side) and did not change in the control group; the erythema index did not change, so this fits pigment-type dark circles only. Small, short, one centre; the microneedling is a device step a cream user does not get. The PR percent is not in the abstract.
- Reference: Yu et al., 2021
Listed in 36 products
- CaffeineLow concern
Studied only mixed with other actives
A combination pad of 3% caffeine + 1% vitamin K in 11 people (Ahmadraji 2015). An RCT of 3% caffeine gel alone (n=34) showed no significant difference from the base for eye puffiness (Amnuaikit 2011). There is no controlled evidence for caffeine alone on vascular dark circles.
- Reference: Ahmadraji and Shatalebi, 2015
- Reference: Amnuaikit et al., 2011
- Reference: Rajabi-Estarabadi et al., 2024
Pregnancy: – · Listed in 1,317 products
- NiacinamideLow concern
Studied at 2-5% (other use)
- Korea approved amount: 2-5%
A limit is the most a product may hold, not the amount that works.
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Bissett et al., 2004
- Reference: Draelos et al., 2006
- Reference: Kimball et al., 2010
- Reference: Korea Ministry of Government Legislation, 2023
- Reference: Brady and Shah-Desai, 2025
Pregnancy: – · Listed in 12,904 products
- All-trans-RetinolModerate concern
Retinol
Studied at 0.075% (other use), 0.4% (other use)
- Korea approved amount: about 0.075%
A limit is the most a product may hold, not the amount that works.
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement. Ye 2022: 33 women used a product with supramolecular retinol plus Acmella oleracea extract around the eyes for 6 weeks with no comparison group; the melanin index fell 13.8% (the retinol percent is not in the abstract, and the authors work at the company that made it). The Hamie 2024 review says trials of retinoids in eye-cream formulas are lacking. No under-eye trial was found for retinal or the retinol esters.
- Reference: Kafi et al., 2007
- Reference: Kikuchi et al., 2009
- Reference: Korea Ministry of Government Legislation, 2023
- Reference: Mitsuishi et al., 2004
- Reference: Ye et al., 2022
- Reference: Hamie et al., 2024
Pregnancy: advised to avoid · Listed in 752 products
- Tetrahexyldecyl ascorbateLow concern
Studied only mixed with other actives
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Fitzpatrick and Rostan, 2002
Pregnancy: – · Listed in 300 products
- Artemisia Absinthium ExtractLow concern
Studied at 20%
Hamdi 2023 (60 enrolled, 45 finished, randomized, double-blind, placebo-controlled): a cream with 20% aqueous wormwood extract (standardized to 1.29 ug/mg artemisinin) was applied under the eyes. After treatment the melanin index was lower than in the placebo group (p=0.01); the redness index did not differ between groups, so this fits pigment-type dark circles only. The two groups already differed in color values at the start (unbalanced baseline), and it is one small trial from one place. The length of use is not in the abstract.
- Reference: Hamdi et al., 2023
Listed in 189 products
- Hyaluronan 130 kDaLow concern
Studied at 0.1% (other use)
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Pavicic et al., 2011
Not in our product list yet
- Hyaluronan 50 kDaLow concern
Studied at 0.1% (other use)
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Pavicic et al., 2011
Not in our product list yet
- N-Acetyl-D-glucosamineLow concern
Acetyl Glucosamine
Studied at 2% (other use)
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Bissett et al., 2007
- Reference: Kimball et al., 2010
Pregnancy: – · Listed in 603 products
- Sodium Acetylated HyaluronateLow concern
Studied at 0.2% (other use), 0.1% (other use)
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Meunier et al., 2022
- Reference: CIR, 2023
Pregnancy: – · Listed in 3,685 products
- Tranexamic acidLow concern
Study dose: not found yet
A candidate for pigment-type dark circles. Human evidence is only intradermal-injection mesotherapy (Iranmanesh 2024, n=18) and an uncontrolled combination (Brady 2025, n=53). No study testing topical tranexamic acid alone under the eyes was found. A limited inference from the melasma evidence (txa, txa3).
- Reference: Iranmanesh et al., 2024
- Reference: Brady and Shah-Desai, 2025
- Reference: Janney et al., 2019
Pregnancy: – · Listed in 767 products
- Ascorbyl glucosideLow concern
Study dose: not found yet
In the same Ohshima 2009 study, the ascorbyl glucoside lotion did not differ significantly from vehicle in erythema index, melanin index or dermal thickness. Rated low as honest evidence that it was tested and showed no effect.
- Reference: Ohshima et al., 2009
Pregnancy: – · Listed in 784 products
- Acetyl hexapeptide-8Moderate concern
Study dose: not found yet
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Wang et al., 2013
- Reference: Henseler, 2023
- Reference: FDA GSRS
Pregnancy: – · Listed in 2,642 products
- ArbutinModerate concern
Study dose: not found yet
A candidate for pigment-type dark circles. The only evidence is an uncontrolled single-arm study (Brady 2025) of a combination containing arbutin, so arbutin's share cannot be sorted out.
- Reference: Brady and Shah-Desai, 2025
- Reference: Korea Ministry of Government Legislation, 2023
Pregnancy: – · Listed in 478 products
- Gentiana Lutea Root ExtractLow concern
Studied at 147 µg/mL
Berthon 2023 (22 people, split-face: one side extract cream, the other placebo, twice a day for 14 days): a gel cream with 147 ug/mL Gentiana lutea root extract (gentiopicroside 65% of dry matter) cut upper-eyelid roughness and the redness of dark circles. The authors work at the company that sells this extract. The abstract does not name the plant part; root comes from that company's ingredient listing. Short and small.
- Reference: Berthon et al., 2023
Listed in 178 products
Copper Tripeptide-1
Study dose: not found yet
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Badenhorst et al., 2016
Pregnancy: – · Listed in 2,137 products
- Hyaluronan 2000 kDaLow concern
Studied at 0.1% (other use)
The definite effective concentration for the periorbital area itself is N/A or limited to a specific combination. Moisturizing, wrinkle and pigment results are not turned into vascular dark-circle improvement.
- Reference: Pavicic et al., 2011
Not in our product list yet
- Salix Alba Bark ExtractLow concern
Studied at 2%
Jeon 2023 (29 people): a cream with 2% Salix alba bark extract for 8 weeks improved the dark-circle measures, skin microcirculation and elasticity. The abstract mentions no comparison group, so placebo effect or season cannot be separated. The cell work in the same paper (more hyaluronic acid, less vascular leak) is not a human trial. One small trial.
- Reference: Jeon et al., 2023
Pregnancy: – · Listed in 945 products
- Palmitoyl tripeptide-1Low concern
Studied only mixed with other actives
Evidence exists only for Haloxyl, an eye-care ingredient mix. In Souza 2013 (62 of 80 finished) the 2% Haloxyl gel group, 3 months, was rated high in the treating physician's satisfaction rating (patient ratings did not differ clearly between groups). There is no untreated group, and the ratings are satisfaction ratings from photos, not a measurement of pigment. The paper does not name the peptides in the mix (it says chrysin, N-hydroxysuccinimide and 'matrikine' peptides); that palmitoyl tripeptide-1 is one of them comes from the raw-material maker's data sheet, which is not peer reviewed. This one ingredient's own share cannot be separated.
- Reference: Souza et al., 2013
Pregnancy: – · Listed in 1,680 products
- Palmitoyl tetrapeptide-7Low concern
Studied only mixed with other actives
Same reason as the palmitoyl tripeptide-1 row. Tested only as part of the Haloxyl mix (Souza 2013, 62 people, no untreated group, satisfaction ratings), and the paper does not name the peptides. That palmitoyl tetrapeptide-7 is in the mix comes from the raw-material maker's data sheet, which is not peer reviewed. This one ingredient's own share cannot be separated.
- Reference: Souza et al., 2013
Pregnancy: – · Listed in 1,254 products
– = no pregnancy research data yet
Which ingredients for dark circles need a doctor?
2 ingredients here are prescription-only or for in-clinic use in the US. We list them so you can look them up; they never go into a PERBETE routine.
Study dose: not found yet
Prescription drug: shown on lookup pages only. Souza 2013 (62 of 80 finished, no untreated or vehicle group, the measure is satisfaction ratings from photos): 2% hydroquinone cream for 3 months. The abstract lists this group among the best, but in the body a gel with an eye-area peptide mix and the thioglycolic stick rated higher, and the authors suspect the 2% strength was too low (for the pigmented type the blinded rater also rated hydroquinone well). Momosawa 2008 (18 patients, no control): 6 weeks of tretinoin gel plus hydroquinone ointment, then a laser, repeated 2-4 times; 83% rated excellent or good, but drugs and laser were used together so hydroquinone's own share cannot be separated. The Souza authors cite earlier reports that high strengths can pigment and cloud the cornea (not something they saw in this trial).
- Reference: Souza et al., 2013
- Reference: Momosawa et al., 2008
Pregnancy: advised to avoid · Listed in 1 product
- PhytonadioneHigh concern
Studied only mixed with other actives
Uncontrolled studies such as past 2% phytonadione + retinol 0.1%. Because it is an EU cosmetic-banned ingredient it is excluded from general periorbital recommendations; shown for historical evidence.
- Reference: Mitsuishi et al., 2004
- Reference: European Union
Listed in 4 products
Which ingredients for dark circles are advised against in pregnancy?
In our data, pregnancy guidance advises avoiding All-trans-Retinol and Hydroquinone. Each ingredient page names the source.
Pregnant or breastfeeding? Check new products with your doctor or midwife.
Get a routine for your skin
A free quiz picks products for your concerns and leaves out ingredients you tell it to avoid. Answer 5 questions, skip the rest.