Tretinoin Under Eyes: What Studies Show, Irritation Risk and Gentler Options

By Peptide Skincare Editorial Team · Published November 4, 2026 · Updated November 1, 2026 · 15 min read

Start with our pillar guide: What Is the Skin Barrier? The Science of Hydration, Ceramides and Occlusives, or browse all Skincare Fundamentals articles.

Some studies of tretinoin on the face have measured results around the eyes; we found none on the eyelid itself. An early study found it improved fine lines but not deep furrows, and irritation is the main problem. We found no study of eye-contact safety, so the decision belongs with your prescriber. Retinaldehyde was better tolerated in one trial.

Evidence: Limited.

Key takeaways

A 6-month study of 15 patients with sun damage found topical tretinoin improved fine periorbital lines but not deep furrows (Tadaki et al., 1993).
Tadaki et al. described irritation from 0.1% tretinoin cream in Japanese patients as "unexpectedly severe" compared with reports in Caucasians (Tadaki et al., 1993).
In a 44-week trial of 125 patients, 0.05% retinoic acid and 0.05% retinaldehyde both reduced crow's feet wrinkle and roughness features at week 18, but retinoic acid caused more irritation and affected compliance in the 135-patient safety population (Creidi et al., 1998).
A 24-week randomized trial found no significant difference between adapalene 0.3% gel and tretinoin 0.05% cream on periorbital wrinkles, with a similar safety profile (Bagatin et al., 2018).
A 2024 review of eye-cream ingredients named a "lack of clinical trials specifically targeting eye cream formulations and periorbital skin" as a key limitation (Hamie et al., 2024).

Can you use tretinoin under your eyes?

Studies have measured results around the eyes, including crow's feet, in people treating facial sun damage; the abstracts do not say how close to the eye the cream went, and we found none on the eyelid itself. Whether it suits your eye area is a question for your prescriber, because irritation can limit use and we found no eye-safety data.

Tretinoin is the prescription form of retinoic acid, the compound that retinol and retinaldehyde are precursors of. Periorbital simply means the skin surrounding the eye. Search results disagree on whether this skin is off-limits, and a recurring reader worry is that the area may peel or sting more easily than the rest of the face.

The research doesn't settle that worry. A 2024 review of eye-cream ingredients said retinoids "show promise" for wrinkles and hyperpigmentation but flagged a "lack of clinical trials specifically targeting eye cream formulations and periorbital skin" (Hamie et al., 2024). The authors declared no conflicts.

This page covers eye-area safety only. For the broader question, see whether tretinoin helps with wrinkles. Our skin barrier guide explains the hydration terms used in these studies.

Evidence: Limited.

What do tretinoin under eyes studies actually measure?

They measure periorbital skin and crow's feet; we found none on the eyelid itself. The results support modest change in fine lines, and one trial's abstract reports a "less pronounced reduction" at week 44, with no comparison point.

The earliest study here treated 15 middle-aged or elderly patients with chronic sun damage, plus 8 with melasma and 3 with xeroderma pigmentosum, using topical tretinoin for 6 months (Tadaki et al., 1993). Fine surface lines in the periorbital region improved significantly, while deep furrows did not. The abstract reports no control group or funding source, so this is weak evidence on its own.

The strongest eye-area data come from a controlled trial of 125 patients with facial photodamage, indexed as randomized (Creidi et al., 1998). It compared 0.05% retinoic acid (n=40), 0.05% retinaldehyde (n=40) and vehicle, a base cream without the active ingredient (n=45). Researchers took silicone molds of the left crow's feet area and analyzed them with optical profilometry, which measures surface relief.

Both active creams significantly reduced wrinkle and roughness features at week 18. At week 44, the abstract reports a "less pronounced reduction" in both active groups, without a size or comparison point, and vehicle showed no significant change at any point. The abstract doesn't report funding. Missing funding details aren't proof of independence.

When tretinoin studies assessed results

Week 18
Crow's feet features reduced
Retinoic acid 0.05% and retinaldehyde 0.05% both showed significant reductions (Creidi).
Week 24
Final head-to-head assessment
Adapalene 0.3% and tretinoin 0.05% did not differ significantly on periorbital wrinkles (Bagatin).
6 months
Fine lines improved, deep furrows did not
This result comes from a study of 15 patients with sun damage (Tadaki).
Week 44
Reduction less pronounced
The abstract reports a 'less pronounced reduction' in both active groups, with no size or comparison point given (Creidi).

Evidence: Limited.

Is irritation the main risk of tretinoin under eyes?

Yes, on current evidence. Irritation is the main reported problem, though no abstract we found gives an irritation rate for the eye area specifically.

In the 1993 study, the authors called irritation from 0.1% tretinoin cream "unexpectedly severe" in Japanese patients compared with reports in Caucasians (Tadaki et al., 1993). That comparison was with earlier reports, not a side-by-side test. The abstract gives no percentage and doesn't say how much of the irritation occurred around the eyes.

The same study measured the stratum corneum, the skin's outer layer. Water content and transepidermal water loss, the water escaping through skin, both rose 1 month after starting tretinoin. Those readings came from the forearm, so they don't tell us how periorbital skin responds.

The Creidi trial offers the clearest practical signal. Among 135 patients in its safety population, retinoic acid caused more local irritation than retinaldehyde and "affected compliance," meaning it interfered with how consistently people used it (Creidi et al., 1998). The authors of a later eye-cream study likewise described retinoids as irritating enough to limit use in this "vulnerable area" (Kaufman et al., 2022).

We found no study of eye contact, dry eye or eyelid dermatitis with tretinoin. As a general precaution from our editors, not trial data, keep retinoids away from the lash line and call a doctor if your eyes sting, swell or stay red. Evidence: Limited.

Which retinoids may be gentler than tretinoin under eyes?

Retinaldehyde has the best tolerance signal, based on 1 head-to-head trial against retinoic acid. Adapalene matched tretinoin on periorbital wrinkles, but its safety profile was similar, so it wasn't shown to be gentler.

Retinaldehyde

Retinaldehyde is a natural precursor of retinoic acid. In the Creidi trial, it was well tolerated throughout and reduced crow's feet features alongside retinoic acid at week 18 (Creidi et al., 1998).

Retinoic acid vs retinaldehyde in a 125-patient crow's feet trial
Feature	Retinoic acid 0.05%	Retinaldehyde 0.05%
Patients in efficacy analysis	40	40
Week 18	Significant reduction in wrinkle and roughness features	Significant reduction in wrinkle and roughness features
Week 44	Less pronounced reduction	Less pronounced reduction
Tolerance (safety population 135)	More local irritation	Well tolerated
Compliance (safety population 135)	Affected	No compliance problem reported in the abstract

Bottom line: Both reduced crow's feet features at week 18, and retinaldehyde was better tolerated.

A randomized, double-blind trial of 40 Korean women compared retinaldehyde 0.1% with 0.05% for 3 months (Kwon et al., 2018). Crow's feet texture improved 13.7% and 12.6%, respectively, with no statistical difference between groups, and both creams were well tolerated. The abstract describes no vehicle group, and it doesn't report funding.

An 8-week, investigator-blinded split-face trial in 23 Korean volunteers tested retinaldehyde in multilamellar vesicles, a layered delivery system, against retinol (Kim et al., 2021). Both sides improved, including crow's feet wrinkle depth. The retinaldehyde side did better on objective measures except dermal density, with no adverse events. The abstract doesn't report funding.

Adapalene

Adapalene is another retinoid. An investigator-blinded, parallel-group randomized trial in Brazil compared adapalene 0.3% gel with tretinoin 0.05% cream for 24 weeks (Bagatin et al., 2018). The 2 did not differ significantly on periorbital wrinkles, and adapalene was non-inferior with a similar safety profile. The abstract gives neither sample size nor funding.

Retinol and other cosmetic retinoids

Retinol has the longest trial here. In a 52-week, double-blind, vehicle-controlled study of 62 people, 0.1% stabilized retinol beat vehicle on photodamage at all timepoints and improved crow's feet fine lines "by 44%" (Randhawa et al., 2015). The abstract gives no matching vehicle figure, so we can't report the size of the difference. It also doesn't report funding, and the product went on the full face, not the eye area alone.

Other studies have clear limits or company ties:

Retinyl propionate plus climbazole: A 16-week randomized split-face trial of 45 women found this combination improved ageing attributes significantly more than 0.1% retinol; the authors concluded benefits were "comparable or greater." It caused significantly less irritation in a separate 5-day patch study. All authors worked at Unilever R&D (Hawkins et al., 2017).
Cosmetic retinol product: In a 12-week open-label study with 23 completers, periocular fine and coarse lines improved. Allergan sponsored it, 4 authors were AbbVie employees who may hold AbbVie stock, and the other 2 received consulting fees from Allergan Aesthetics (Goberdhan et al., 2022).
Retinoid and lactic acid eye cream: In a 12-week open-label study with 26 completers, the nightly-use group (n=16) improved 33% from baseline on lines and wrinkles, with no related adverse events (Kaufman et al., 2022). The abstract doesn't report funding.

Key retinoid studies with eye-area results

Tadaki et al., Tohoku J Exp Med, 1993	15 with sun damage (26 total)	6 months	Topical tretinoin; 0.1% cream named for irritation	Fine periorbital lines improved; deep furrows did not
Creidi et al., J Am Acad Dermatol, 1998	125 in efficacy analysis; 135 in safety population	44 weeks	0.05% retinoic acid vs 0.05% retinaldehyde vs vehicle	Both actives reduced features at week 18; abstract reports a 'less pronounced reduction' at week 44
Bagatin et al., Eur J Dermatol, 2018	Not reported in abstract	24 weeks	Adapalene 0.3% gel vs tretinoin 0.05% cream	No significant difference on periorbital wrinkles; similar safety
Kwon et al., J Cosmet Dermatol, 2018	40 women	3 months	Retinaldehyde 0.1% vs 0.05%	Crow's feet texture improved 13.7% and 12.6%; no group difference
Kim et al., J Cosmet Dermatol, 2021	23	8 weeks	Vesicle retinaldehyde 0.05%/0.1% vs retinol 0.05%/0.1%	Retinaldehyde better on objective measures except dermal density
Randhawa et al., J Drugs Dermatol, 2015	62	52 weeks	0.1% stabilized retinol vs vehicle	Significant vs vehicle; crow's feet fine lines improved by 44%

Tadaki et al., Tohoku J Exp Med, 1993Sample size: 15 with sun damage (26 total) · Length: 6 months · Dose: Topical tretinoin; 0.1% cream named for irritation

Result: Fine periorbital lines improved; deep furrows did not

No control group reported in the abstract. Funding not reported. Evidence: Early / preliminary.

Creidi et al., J Am Acad Dermatol, 1998Sample size: 125 in efficacy analysis; 135 in safety population · Length: 44 weeks · Dose: 0.05% retinoic acid vs 0.05% retinaldehyde vs vehicle

Result: Both actives reduced features at week 18; abstract reports a 'less pronounced reduction' at week 44

Controlled trial. Retinoic acid caused more irritation and affected compliance (safety population 135). Funding not reported in abstract. Evidence: Limited.

Bagatin et al., Eur J Dermatol, 2018Sample size: Not reported in abstract · Length: 24 weeks · Dose: Adapalene 0.3% gel vs tretinoin 0.05% cream

Result: No significant difference on periorbital wrinkles; similar safety

Investigator-blinded randomized trial. Funding not reported in abstract. Evidence: Limited.

Kwon et al., J Cosmet Dermatol, 2018Sample size: 40 women · Length: 3 months · Dose: Retinaldehyde 0.1% vs 0.05%

Result: Crow's feet texture improved 13.7% and 12.6%; no group difference

Double-blind randomized trial; abstract describes no vehicle group. Funding not reported in abstract. Evidence: Limited.

Kim et al., J Cosmet Dermatol, 2021Sample size: 23 · Length: 8 weeks · Dose: Vesicle retinaldehyde 0.05%/0.1% vs retinol 0.05%/0.1%

Result: Retinaldehyde better on objective measures except dermal density

Investigator-blinded split-face randomized trial. No adverse events. Funding not reported in abstract. Evidence: Early / preliminary.

Randhawa et al., J Drugs Dermatol, 2015Sample size: 62 · Length: 52 weeks · Dose: 0.1% stabilized retinol vs vehicle

Result: Significant vs vehicle; crow's feet fine lines improved by 44%

Double-blind, vehicle-controlled trial. No vehicle figure or funding reported in abstract. Evidence: Limited.

Concentrations are study arms, not recommendations. Ask your prescriber about product, strength and schedule.

Evidence: Limited.

Does a retinol eye cream work better than tretinoin?

We don't know, because we found no study comparing a retinol eye cream directly with tretinoin. Newer eye-area trials use retinol as the comparator or, in their abstracts, describe no comparator.

A randomized, double-blind, split-face study compared a 0.1% PDRN eye cream with 0.1% retinol over 28 days (Ye et al., 2026). PDRN, or polydeoxyribonucleotide, is a DNA-derived ingredient. The cream achieved "approximately two-fold greater" periocular improvements than retinol, with good tolerability. The abstract gives no sample size, and 3 of the 5 authors list Shanghai biotech companies, though the authors declared no conflicts.

An open-label, 28-day study of a blend containing retinal, another name for retinaldehyde (the abstract's lab work used 0.1%), reported a 43.2% drop in wrinkle count and fewer crow's feet wrinkles in every participant (Brown et al., 2023). The abstract gives no sample size and describes no comparison group. Most authors were ISDIN employees, 1 was a former employee and paid consultant, and ISDIN makes the product.

Neither result tells you how these products compare with tretinoin. For the question of frequency, see whether you can use retinol every night. If you're weighing a non-retinoid option, read peptides vs retinol. Evidence: Early / preliminary.

Who should ask a doctor before using tretinoin under eyes?

Anyone who is pregnant or trying to conceive, or who has eczema or sensitive eyelids, should get medical advice first. These are general precautions from our editors, not trial data, because the abstracts of the eye-area studies we found don't describe those groups.

Our only pregnancy-related trial note concerns an oral drug, not topical tretinoin. In a 3-month randomized trial of 32 menopausal or sterilized women, oral isotretinoin "seemed safe but not effective" for photoaging compared with moisturizer and sunscreen, and the authors urged caution for women who could become pregnant (Bagatin et al., 2010). That finding can't be applied directly to a cream, so ask your doctor.

For eczema, eyelid dermatitis or a history of reacting to retinoids, we found no trial data. Any caution here is a general precaution from our editors. Your prescriber can review your skin, your eye health and the other products you use. Evidence: Limited for the oral isotretinoin trial; no direct data for topical tretinoin in these groups.

What other under-eye concerns are separate from tretinoin under eyes?

Milia and hollows are different problems, and the abstracts of the tretinoin studies above don't report either one.

Small white bumps: See our guide to milia around the eyes.
Lost volume or shadowing: See our guide to under-eye hollows.
Recent procedures: See our separate page on retinoid timing around injections.

Where the evidence stops

No eyelid studies found. The research we found covers crow's feet and periorbital areas; we found none on eyelid skin.
No eye-safety data found. We found nothing on eye contact, dry eye or eyelid dermatitis with tretinoin.
Few tretinoin trials. The key studies date from 1993, 1998 and 2018, and their abstracts don't report funding.
Company ties are common. Many newer eye-area studies are linked to cosmetic manufacturers.
Long-term results are uncertain. One trial's abstract reports only a "less pronounced reduction" at week 44, with no size or comparison point.
Skin tone data are thin. Only 1 cohort in the abstracts, in an open-label eye cream study, included Fitzpatrick skin types up to VI.

FAQ

Can I put tretinoin under my eyes?

Studies have measured results around the eyes, including crow's feet, in people treating facial sun damage, though the abstracts don't say how close to the eye the cream went, and an early study found improvement in fine lines. We found no study of the eyelid itself, so the decision belongs with your prescriber.

Is tretinoin too strong for the eye area?

It can be irritating. One study described irritation from 0.1% tretinoin cream as unexpectedly severe in Japanese patients compared with reports in Caucasians, and in a trial with 135 patients in its safety analysis, 0.05% retinoic acid, the active form sold as tretinoin, caused more local irritation than retinaldehyde and affected compliance. These abstracts don't say where irritation occurred.

Will tretinoin fix deep crow's feet?

Probably not, based on limited data. A 6-month study of 15 patients with sun damage found significant improvement in fine periorbital lines but not in deep furrows; the abstract doesn't say where the deep furrows were.

Is retinaldehyde or adapalene gentler around the eyes?

Retinaldehyde caused less local irritation than retinoic acid in a trial with 135 patients in its safety analysis. Adapalene 0.3% gel had a similar safety profile to tretinoin 0.05% cream in a 24-week randomized trial, so that trial matched it on results but did not show it was gentler. Neither abstract says where any irritation occurred.

Does retinol eye cream work?

Some trials show improvement, but the evidence is modest. A 52-week trial of 62 people found a full-face 0.1% stabilized retinol product beat vehicle on photodamage, but that wasn't an eye cream, and many eye-area retinoid studies are open-label or tied to the companies that make the products.

How long does it take to see results from tretinoin around the eyes?

Tretinoin studies assessed results at 18 weeks, 24 weeks and 6 months, and their abstracts do not say when change first appeared. In a 125-patient trial, the abstract says a less pronounced reduction was seen at week 44, without giving its size or a comparison point.

What if tretinoin gets in my eye?

We found no study of this. As a general precaution, not trial data, our editors suggest rinsing with water and contacting your doctor if stinging, swelling or redness persists.

Can I use tretinoin under my eyes while pregnant?

Ask your doctor. Retinoids carry pregnancy cautions, and the study abstracts we found do not address topical tretinoin in pregnancy.

References

Tadaki T, et al. The effect of topical tretinoin on the photodamaged skin of the Japanese. Tohoku J Exp Med. 1993. Clinical study, 15 patients with chronic sun damage plus 8 with melasma and 3 with xeroderma pigmentosum, 6 months; no control group or funding reported in abstract. https://pubmed.ncbi.nlm.nih.gov/8236244/
Creidi P, et al. Profilometric evaluation of photodamage after topical retinaldehyde and retinoic acid treatment. J Am Acad Dermatol. 1998. Controlled trial, indexed as randomized, n=125 (safety population 135), 44 weeks; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/9843009/
Bagatin E, et al. Comparable efficacy of adapalene 0.3% gel and tretinoin 0.05% cream as treatment for cutaneous photoaging. Eur J Dermatol. 2018. Investigator-blinded RCT, n not reported in abstract, 24 weeks; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/30105991/
Hamie H, et al. A review of the efficacy of popular eye cream ingredients. Int J Womens Dermatol. 2024. Literature review; no conflicts declared. https://pubmed.ncbi.nlm.nih.gov/38873621/
Kwon HS, et al. Efficacy and safety of retinaldehyde 0.1% and 0.05% creams used to treat photoaged skin: A randomized double-blind controlled trial. J Cosmet Dermatol. 2018. RCT, n=40 women, 3 months; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/29663701/
Kim J, et al. The efficacy and safety of multilamellar vesicle containing retinaldehyde: A double-blinded, randomized, split-face controlled study. J Cosmet Dermatol. 2021. Split-face RCT, n=23, 8 weeks; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/33569865/
Randhawa M, et al. One-year topical stabilized retinol treatment improves photodamaged skin in a double-blind, vehicle-controlled trial. J Drugs Dermatol. 2015. Vehicle-controlled trial, n=62, 52 weeks, plus a 12-person forearm substudy; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/25738849/
Hawkins S, et al. Retinyl propionate and climbazole combination demonstrates clinical improvement to the appearance of hyperpigmentation and deep wrinkling with minimal irritation. Int J Cosmet Sci. 2017. Split-face RCT, n=45, 16 weeks; all authors at Unilever R&D. https://pubmed.ncbi.nlm.nih.gov/28733999/
Goberdhan LT, et al. Assessing changes in facial skin quality using noninvasive in vivo clinical skin imaging techniques after use of a topical retinoid product in subjects with moderate-to-severe photodamage. Skin Res Technol. 2022. Open-label study, 23 completed, 12 weeks; sponsored by Allergan; 4 of 6 authors AbbVie employees, other 2 received Allergan Aesthetics consulting fees. https://pubmed.ncbi.nlm.nih.gov/35691012/
Kaufman J, et al. Efficacy and Tolerability of a Retinoid Eye Cream for Fine to Moderate Wrinkles of the Periorbital Region. J Drugs Dermatol. 2022. Open-label study, 26 completed, 12 weeks; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/36074511/
Ye R, et al. Topical medium-length PDRN enhances dermal extracellular matrix repair in photodamaged skin via PI3K-Akt/TGF-β-regulated pathways. PLoS One. 2026. Laboratory, ex vivo and clinical work including a split-face RCT, n not reported in abstract, 28 days; 3 of 5 authors (Ye, Du, Hu) list Shanghai biotech companies (Inertia Shanghai Biotechnology, DermaHealth Shanghai Biotechnology); no conflicts declared. https://pubmed.ncbi.nlm.nih.gov/42430369/
Brown A, et al. Natural Retinol Analogs Potentiate the Effects of Retinal on Aged and Photodamaged Skin: Results from In Vitro to Clinical Studies. Dermatol Ther (Heidelb). 2023. Laboratory and ex vivo work plus an open-label clinical study, n not reported in abstract, 28 days; ISDIN employees and a paid consultant, and ISDIN makes the product. https://pubmed.ncbi.nlm.nih.gov/37615835/
Bagatin E, et al. A randomized and controlled trial about the use of oral isotretinoin for photoaging. Int J Dermatol. 2010. RCT, n=32 women, 3 months; funding not reported in abstract. https://pubmed.ncbi.nlm.nih.gov/20465648/

Disclosure: This article is for general information only. It is not medical advice. Peptide Skincare has no affiliate links in this article and received no payment from any manufacturer mentioned. Tretinoin is a prescription drug. This article is not medical advice; ask your prescriber before using any retinoid near the eyes, and see a doctor if your eyes sting, swell or stay red.

For educational purposes only; not medical advice. See our disclaimer and research method.