Most wrinkle creams are treating the wrong kind of line.

Sunscreen is the only anti-wrinkle intervention proven in a randomised trial to slow aging itself. In the Nambour trial, adults under 55 assigned to daily broad-spectrum SPF 15+ showed no detectable increase in skin aging over 4.5 years on silicone skin casts — 24% less aging than those using sunscreen when they felt like it (Hughes 2013). Photodamage is present in 80–90% of fair-skinned adults, and UV is the largest single cause of the collagen breakdown that makes wrinkles (review).

Practicalities: SPF 30–50, broad-spectrum (UVA is the wrinkle wavelength, and the EU's UVA seal or PA rating matters more than the SPF number), a quarter-teaspoon for the face, every day including winter and indoors near windows, reapplied when outdoors. Mineral or chemical filters both work; the one you will actually wear is the right one. Sunscreen does not undo existing lines — it stops the next ones.

Tretinoin is the reference anti-wrinkle molecule because it is the one that has been tested properly. Two 24-week randomised, vehicle-controlled trials in 533 people found 0.05% emollient cream reduced fine wrinkling, roughness, mottled pigmentation and laxity (overview); 48-week multicentre trials confirmed continued improvement (48-week studies); even 0.02% worked in two double-blind studies (0.02% cream); and a systematic review and meta-analysis of the randomised trials pools the effect (meta-analysis). Biopsies show the mechanism — thicker epidermis, new collagen in the upper dermis, reorganised elastic fibres.

The catch is tolerance: dryness, redness and flaking for the first six to eight weeks, which is why most people quit. Start 0.025% two or three nights a week on dry skin over a moisturiser, build to nightly, and expect visible change at three to four months and the full effect at a year. Not in pregnancy; always with sunscreen. Adapalene 0.3% was non-inferior to tretinoin 0.05% in one comparison and is gentler (comparison).

Oral collagen peptides are absorbed as small fragments that appear to signal fibroblasts. A 2019 systematic review of 11 studies in 805 people found that oral collagen increased skin elasticity, hydration and dermal collagen density — "Preliminary results are promising" (2019 systematic review) — and a 2025 meta-analysis of 23 randomised trials found significant gains in hydration, elasticity and wrinkles across all of them (2025 meta-analysis). Its conclusion that there is "no clinical evidence" goes too far: it rests on the few trials without pharmaceutical-company funding, two of which gave only 1–2 g a day, and the clearest negative trial measured forearm skin in women aged 60 to 93 with fragile, thinning skin (our reading of it). Most trials are funded by the makers, and the effects are the kind you measure with an instrument rather than see across a room.

Take about 10 g a day with vitamin C and other skin nutrients. The skin may also need a local signal to build collagen where the lines are: a nightly retinoid or a collagen-stimulating treatment. Collagen alone at a low dose may not be enough. Our collagen guide grades the forms and doses; the honest summary is "real, small, and best on top of a retinoid".

Retinol has to be converted twice in the skin to become tretinoin, so it is weaker gram for gram but far better tolerated. The evidence is real: in the classic study, 0.4% retinol applied to one arm of older adults up to three times a week for 24 weeks reduced fine wrinkles and increased collagen and glycosaminoglycan production on biopsy (Kafi 2007); 0.3% retinol improved multiple photoaging measures in a later study (2022 study); and in a randomised split-face comparison, a well-formulated retinol regimen approached the tretinoin regimen on wrinkle scores with less irritation (split-face study). Retinaldehyde, one step closer to tretinoin, matched a course of glycolic peels in a randomised study (2018 trial).

What to buy: a stated concentration (retinol up to 0.3%, the EU limit for face products, or 0.05–0.1% retinaldehyde), airless opaque packaging (retinoids oxidise), and patience — the trials ran 24 weeks. Most "retinol" products list no percentage and contain a token amount.

Vitamin C is a cofactor for collagen synthesis and an antioxidant that mops up the free radicals UV generates. Topically, the evidence is small but controlled: in a six-month double-blind, randomised trial, a 5% cream significantly reduced deep furrows and increased the density of skin microrelief on the treated side (Humbert 2003); an earlier 12-week double-blind half-face study found clinically visible improvement in wrinkling with biopsy evidence of new collagen (Fitzpatrick 2002). Combined with vitamin E and ferulic acid it also measurably reduces UV damage under sunscreen, which is its most defensible job.

The practical problem is chemistry: L-ascorbic acid oxidises in weeks once opened and goes yellow-brown. Buy 10–20% in opaque, airless packaging, keep it cool, and discard when it darkens. Derivatives are stabler and weaker.

Niacinamide (vitamin B3) supports the skin barrier, calms inflammation and reduces pigment transfer. In the best study, 50 women applied 5% niacinamide to one side of the face and vehicle to the other, twice daily for 12 weeks, double-blind: the niacinamide side showed significant reductions in fine lines and wrinkles, hyperpigmented spots, red blotchiness and sallowness, and improved elasticity on cutometry (Bissett 2005). A cosmetic regimen of niacinamide, peptides and retinyl propionate was comparable to 0.02% tretinoin for wrinkles in a randomised comparison — a fair statement of its ceiling (regimen trial). The 4–5% concentration in the trials is the useful dose; the 10% versions add irritation, not effect.

Dehydrated skin creases more, and hydrating it uncreases it — for as long as it stays hydrated. A serum with five forms of hyaluronic acid produced statistically significant reductions in fine and coarse lines and roughness immediately after application, with cumulative improvement over weeks of use (JAAD, 2016); a 2023 review of topical hyaluronic acid reaches the same conclusion about hydration and the appearance of aging (review). Topical hyaluronic acid does not penetrate to the dermis and does not build anything; it holds water in the surface, which is exactly what makes a fine line look shallower at 9 a.m.

Use it for what it does: instant softening, a comfortable buffer under tretinoin, and better barrier function in dry or menopausal skin. Any moisturiser with glycerin and ceramides does most of the same — see the ceramides guide.

Smoking is the cleanest case: among 79 pairs of identical twins discordant for smoking, the smoker scored worse for upper-lip lines, lower-lid bags, nasolabial folds and jowls, and five years of smoking difference was visible on the face (twin study); the larger Guyuron twin series added sun, alcohol and weight swings to the list (Guyuron 2009). Sugar ages skin by glycation — sugars cross-link collagen and elastin into stiff, brittle fibres — which is well established mechanistically and supported by observational data, not trials (review). Poor sleepers showed more intrinsic aging signs and slower barrier recovery in a 60-woman study (Oyetakin-White 2015).

None of this reverses a line. All of it decides how many you make in the next decade, at zero cost.

Alpha-hydroxy acids loosen the bonds between dead surface cells, which smooths texture and brightens within weeks; at higher concentrations and over months they modestly thicken the epidermis. The controlled evidence is one good trial: 74 women aged 40–70 used 8% glycolic acid, 8% lactic acid or vehicle for 22 weeks, and the proportion improving by at least one grade of photodamage was 76%, 71% and 40% respectively (Stiller 1996). A daily 5% glycolic formulation showed only trends toward improvement in a smaller double-blind study (1998 trial).

Useful for texture and dullness, less so for a wrinkle you can feel. They increase sun sensitivity — sunscreen is non-negotiable — and they compete with retinoids for the same irritation budget, so alternate nights rather than layering. Our peel guide covers the in-clinic strengths.

Bakuchiol is a plant compound with retinoid-like gene effects and none of the retinoid chemistry, which is why it is marketed as "retinol for sensitive skin". The evidence is one well-designed trial: 44 people randomised to 0.5% bakuchiol twice daily or 0.5% retinol nightly for 12 weeks, double-blind, with computer-analysed photographs. Both reduced wrinkle surface area and pigmentation with no statistical difference between them, and the retinol group reported more scaling and stinging (Dhaliwal 2019, BJD). One trial against a modest comparator makes it emerging, not proven — but it is the only "natural retinol" with a randomised trial behind it, and it can be used in pregnancy, where retinoids cannot.

Peptides are fragments of collagen or nerve-signalling proteins that, in cell culture, nudge fibroblasts to make matrix or dampen the muscle signal. The clinical evidence is thin and mostly manufacturer-run. Palmitoyl pentapeptide (Matrixyl) was part of a niacinamide-and-retinyl-propionate regimen that matched 0.02% tretinoin for wrinkle reduction in a randomised comparison (regimen trial), and a multi-peptide eye serum improved crow's feet measures in a 2023 controlled study (2023 trial). Acetyl hexapeptide-3 (argireline), sold as "Botox in a bottle", relaxes muscle in a dish; it has never been compared with toxin in a controlled trial, and a topical cannot reach the muscle that makes a frown line.

Verdict: fine in a moisturiser you like, priced as if they were prescriptions. If a peptide product costs more than tretinoin and sunscreen together, the money is in the wrong place.

Growth-factor serums contain signalling proteins harvested from cultured cells and, in a few controlled trials, they do something: a cosmetic growth-factor serum produced significantly greater reductions in fine and coarse lines than placebo at 4, 8 and 12 weeks (JAAD, 2018), and a 2023 systematic review found most studies positive but small, short and often uncontrolled (Quinlan 2023). "Stem-cell" creams contain no living cells; the better ones use the conditioned medium those cells grew in, for which a meta-analysis of randomised trials — mostly as an add-on to lasers and needling — found reductions in wrinkles and pigment (2022 meta-analysis). Large proteins penetrate intact skin poorly, which is why the strongest data come from use after resurfacing. At €100–300 a bottle, the price is far ahead of the evidence.

Red (630–660 nm) and near-infrared light stimulate fibroblast metabolism in a dish, and in people the controlled evidence is real but modest. The reference trial randomised 113 of 136 people to red/near-infrared light twice weekly for 30 sessions and compared them with 23 untreated controls: treated skin showed significantly improved complexion, roughness and collagen density on ultrasound (Wunsch & Matuschka 2014); a 2023 meta-analysis of LED skin therapy found consistent skin-rejuvenation benefits across small trials with heterogeneous devices and doses (2023 meta-analysis). Consumer masks emit a fraction of the irradiance used in trials and there is no head-to-head against a retinoid. Harmless, pleasant, and the least cost-effective item on this list. Our red-light guide covers doses.

Silicone patches hold moisture against the skin and physically prevent the crease from forming while you sleep, so a forehead or chest line looks softer for a short while after removal — the same mechanism as a hydrating serum, plus splinting; on the forearm, the water a silicone sheet adds to the skin is lost quickly once it comes off (Suetak 2000). We found no trial of a silicone wrinkle patch in PubMed (searched October 2026). Adhesive tape pads have been tested: on crow's feet, a placebo-controlled study found they smoothed the skin "in the first hour after use" and were "tolerable for most users" (Mazzarello 2018), and on forehead and frown lines four weeks of nightly use gave "no statistical benefit" on blinded rating (Ryan 2009). No study has shown an effect that outlasts the first hour after removal, and neither product relaxes the muscle. Cheap experiments for sleep creases and chest lines; not treatments.

The study everyone cites enrolled 27 women aged 40–65 in a 20-week programme of 30-minute daily facial exercises; the 16 who finished were rated as having fuller upper and lower cheeks, and blinded estimated age fell from 50.8 to 48.1 years (Alam 2018, JAMA Dermatology). It had no control group, measured fullness rather than lines, and the effect could plausibly be hypertrophy of the cheek muscles. For wrinkles specifically the logic runs the other way: expression lines are made by repeated contraction, and dermatologists who inject toxin spend their days stopping exactly the movements face yoga prescribes. Harmless for cheeks; unproven for anything else.

Botulinum toxin blocks the nerve signal to the small muscles that fold the skin, and for the lines they make it is the treatment. In the pivotal placebo-controlled glabellar trial, 20 units across five sites produced a responder rate at day 30 of 77% by physician rating at maximal frown and 89% by patient assessment, against near-zero for placebo, with adverse events no different from placebo (Carruthers 2003). For crow's feet the phase 3 responder rate was 55% (investigator) versus 3% for placebo (phase 3); for forehead lines, 46–53% improved by two grades versus 0.6% (review); a combined upper-face protocol keeps patients satisfied for up to six months (2022 study). Newer toxins (abo-, inco-, pra-, dax-) are equivalent or slightly longer-lasting.

What it does not do: it does not fill a static groove that has already etched in — that softens over repeated cycles as the dermis recovers, or needs resurfacing; and it does nothing for folds or lip volume. Dose and placement decide the look: a light dose keeps movement, a heavy one freezes. Effect starts at day 3–5, peaks at two weeks and fades at three to four months, with no cumulative harm from stopping. Our 40s guide covers the toxin-and-filler decade in context.

Folds are volume problems, and hyaluronic-acid gel replaces volume. The evidence base is unusually good because regulators demand it: in two randomised split-face pivotal trials, responder rates on the Wrinkle Severity Rating Scale (at least one grade better) were 81% at two weeks and 56% at 26 weeks for both the test gel and the Restylane comparator (two RCTs); in a 12-month randomised comparison of two gels, about 80% of subjects were still responders at a year (12-month trial); a 52-week split-face study found the same (52-week study). MRI studies show the gel persists far longer than the "six months" folklore, which is the argument against topping up on a schedule (MRI review).

The skilled version treats the cause — the deflated cheek — rather than the fold, which is why an injector who reaches for the cheekbone when you ask about the nasolabial line is usually the better one. Risks are rare but real (vascular occlusion, blindness) and covered in the safety drawer; the gel dissolves on demand with hyaluronidase. Our filler guide goes zone by zone.

Soft, low-viscosity hyaluronic-acid gels placed superficially or in the lip fill the deepest perioral grooves and add the small volume whose loss slackens the skin above the lip. The pivotal trial for Volbella in the lips and perioral area showed responder rates sustained to a year (Volbella pivotal trial). Microdroplet "skin boosters" (Restylane Vital) are a different idea — hydration and roughness rather than filling — and in a randomised, open-label, single-centre study of two against three initial sessions, 72–79% of patients rated themselves improved at three months and investigators rated 92–96% improved (randomised study); with Juvéderm Volite, a similar skin-quality gel, the skin-roughness responder rate fell from 96% at month one to 35% by month six in a single-arm study (responder data). Profhilo, the "bioremodelling" injectable, has thinner evidence than its marketing; the filler guide grades it separately.

The trade-off in the lip zone is lumps and visible product in thin skin, which is why the best results combine a small volume of thin gel with a few units of toxin and resurfacing, rather than more filler.

Poly-L-lactic acid microspheres are injected as a suspension and, over two to three months, provoke fibroblasts to lay down new collagen around them, so the volume that appears is the patient's own. In the randomised pivotal comparison against human collagen filler, nasolabial-fold correction was significantly better from month 3 and persisted to 25 months (FDA summary); a 2026 multicentre, double-blinded, randomised trial in 252 people found 92.4% of nasolabial folds still corrected at 48 weeks against 59.3% with hyaluronic acid (2026 trial). Because the effect builds slowly it never looks "done"; because it is collagen it lasts. The price of that is patience, the need for meticulous dilution and massage to avoid nodules, and no hyaluronidase escape hatch if you dislike it.

Calcium hydroxylapatite is a suspension of mineral microspheres in gel; it fills immediately and stimulates collagen as the gel is absorbed, so results outlast the product. In a 60-person split-face trial against a hyaluronic filler, 79% of the calcium hydroxylapatite folds were still improved at 12 months against 43% (2008 split-face trial), and against human collagen in 117 people 79% had the better result on the calcium hydroxylapatite side at six months, with less product (2007 trial); the Radiesse guide grades it area by area. It cannot be dissolved, must not go in the lips, and diluted ("hyperdilute") it doubles as a skin-quality treatment for the cheeks and neck. A good choice for deep folds in a heavier face; a poor one for a first-time patient who might want to undo it.

Platelet-rich plasma injected into or needled into the face delivers growth factors that, in three randomised split-face trials, improved skin texture and fine wrinkles modestly over a few months — with the clearest effects when combined with microneedling or laser rather than alone (three split-face trials). Preparation systems, platelet concentrations and injection depths vary wildly, which keeps the tier at emerging. It is a reasonable add-on for someone already having needling, and a poor stand-alone purchase; the PRP guide covers the systems and the polynucleotide and exosome cousins.

The idea is sound — stop the folding before the crease etches — and the evidence is a single case report: identical twins, one treated with toxin two to three times a year for 13 years, the other twice in total, with the treated twin showing no forehead or glabellar lines at rest and the untreated twin visible ones (Binder's twins). One pair is not a trial. The sensible reading: once a line appears with expression and is starting to persist at rest — typically late twenties to thirties — a light dose is prevention with a plausible mechanism. Injecting a face that has no lines at rest is a subscription, not a treatment. Our 30s guide takes the prejuvenation industry apart in detail.