TL;DR: There is exactly one well-cited head-to-head trial of bakuchiol versus retinol in humans, Dhaliwal 2019, with 44 participants and 12 weeks of use. It found similar improvement in wrinkling and pigmentation, and lower scaling and stinging with bakuchiol. That is the entire dataset most marketing leans on. The trial did not measure dermal collagen, did not compare bakuchiol to tretinoin, did not stratify by skin type, and was not designed to detect long-term differences. Bakuchiol behaves like a gentle retinoid-adjacent ingredient. It is not a replacement for prescription retinoids in the way the bottles imply.
A reader sent me a screenshot last month of a brand caption that said bakuchiol is “as effective as retinol, without the irritation.” She wanted to know if she could swap her 0.05% tretinoin for a bakuchiol serum during pregnancy. The honest answer required walking back almost every claim in the caption.
I have read the bakuchiol literature twice, the first time when the ingredient went viral around 2019 and again last year when a friend formulating a clean-beauty serum asked me to vet the references. There is less data than the marketing implies. The data that exists is mostly favorable for bakuchiol in a narrow sense, but the comparative framing keeps getting stretched past what the studies measured.
What the studies actually show
The paper everyone cites is Dhaliwal 2019 in the British Journal of Dermatology (PMID: 29947134). It is a 12-week prospective, randomized, double-blind split-face style assessment of 0.5% bakuchiol cream twice daily versus 0.5% retinol cream once daily. Forty-four participants completed. Photo evaluation by a blinded dermatologist showed both groups improved on wrinkle area and hyperpigmentation, with no statistically significant difference between the two arms.
That last clause is what gets quoted. What gets left out is the rest of the methods section.
The retinol arm reported more scaling and stinging. That is the secondary finding, and it is real. Bakuchiol was better tolerated. The trial used patient-reported tolerability and a clinical visual scale, not transepidermal water loss or barrier biomarkers, so the tolerability gap is descriptive, not mechanistic.
Sample size matters here. Forty-four participants split into two arms gives you roughly 22 per group. With that sample you can detect medium-to-large effect differences. You cannot detect small ones. The conclusion “no significant difference” with n=22 per arm is not the same as “equivalent.” Statisticians call this a power problem. The trial was not pre-registered as a non-inferiority study, and Dhaliwal et al. were appropriately cautious in the discussion, noting the need for “larger studies.” That caution disappears in marketing copy.
The earlier mechanistic work by Chaudhuri and Bojanowski 2014 (PMID: 24471735) used gene expression profiling on human skin biopsies and reported that bakuchiol upregulates a similar set of collagen-related genes (types I, III, IV) and downregulates matrix metalloproteinases in a pattern that overlaps with retinol. This is the basis for calling bakuchiol “retinoid-like.” It is a reasonable comparison at the gene-expression level. It is not a comparison at the receptor level, because bakuchiol does not appear to bind retinoic acid receptors in any meaningful way.
Bluemke et al. 2022 (PMID: 35514228) is the most thorough mechanistic follow-up, looking at multiple aging pathways in vitro and in a clinical arm. The clinical arm in that paper was open-label, not blinded against retinol. It supports bakuchiol’s standalone effects on fine lines and elasticity. It does not extend the comparative claim.
What the studies do not show
Bakuchiol has not been compared head-to-head with tretinoin in any peer-reviewed trial I can find. The Dhaliwal arm used 0.5% retinol, which is a mid-strength cosmetic retinol, not a prescription retinoid. Prescription tretinoin at 0.025% or 0.05% has a substantially larger effect-size literature in photoaging (Mukherjee et al. 2006, PMID: 18046911), including dermal-level changes on biopsy. Comparing bakuchiol to 0.5% retinol and then implying parity with tretinoin is a category error.
No long-term study exists. Twelve weeks is the longest published comparative window. Retinoid clinical trials have run 24 to 52 weeks for a reason: collagen remodeling is slow and the gap between groups often widens with time.
Skin type stratification is largely absent. The Dhaliwal sample was predominantly lighter Fitzpatrick types. Whether bakuchiol behaves identically on melanin-rich skin in terms of pigmentation outcomes is an open question.
There is no published data on bakuchiol during pregnancy. The marketing implication that “natural means pregnancy-safe” is not how pharmacology works. Bakuchiol is a meroterpene phenol from Psoralea corylifolia, a plant in traditional Ayurvedic and Chinese medicine with its own toxicology questions at higher doses orally. The topical safety record at cosmetic concentrations looks reasonable, but my OB and dermatologist friends are not telling pregnant patients it has been studied for that use, because it has not been.
The contrarian read
I think bakuchiol is a useful ingredient. I keep one in my own rotation for the nights I want barrier-friendly antioxidant activity without a retinoid pulse. What I do not believe is the framing that it is an equivalent swap for tretinoin or even for 0.5% retinol with a longer use horizon.
The marketing pitch sits on two convenient facts. First, the one head-to-head trial showed numerical parity at 12 weeks. Second, bakuchiol stings less. Both are true. Neither establishes that bakuchiol matches a prescription retinoid for photoaging or acne over the timescales that matter.
There is a quieter possibility the literature gestures at: bakuchiol may be most useful as a retinoid companion rather than a substitute. Some formulators pair low-dose retinol with bakuchiol to extend tolerability. Draelos 2020 (PMID: 33346519) is one of the few papers exploring combination logic, and the rationale makes sense biologically even where the data is thin.
The other underdiscussed point is concentration. Most clinical work uses 0.5% bakuchiol. A lot of finished products list bakuchiol on the front of the bottle at concentrations the brand never confirms in writing. I have seen marketing decks where the active concentration is closer to 0.05% than 0.5%, which is the same trick clean-beauty brands have run on niacinamide and on vitamin C for years. The ingredient is in the bottle. The dose is not.
If you cannot get a concentration disclosure from the brand, treat the claim with skepticism. A “bakuchiol serum” without a stated percentage is a marketing artifact more than a formulation specification. The same applies to retinol, where 0.1% and 1% products carry the same shelf label but produce very different clinical effects.
What I would tell my past self
When I first read Dhaliwal in 2019 I was excited the way most skincare readers were. A plant molecule that performs like retinol without the peel-week irritation is a great story. Five years later the story has not been replicated at scale, the sample sizes remain small, and the comparative dataset has not really grown. That is a signal.
If you cannot tolerate retinoids, bakuchiol is a reasonable maintenance ingredient. If you are using a retinoid that works, do not swap it for bakuchiol because a caption told you they are equivalent. The available evidence does not carry that claim.
Frequently asked
Is bakuchiol pregnancy-safe?
There is no human data on topical bakuchiol during pregnancy. The assumption that “natural” means safe is not pharmacologically sound. I would discuss it with your OB rather than rely on brand claims.
Can I use bakuchiol with retinol?
Yes, and there is some preliminary rationale for the combination from Draelos 2020. Most well-formulated combination products use a low retinol percentage with bakuchiol as a co-active.
Does bakuchiol work for acne?
There is no comparative trial of bakuchiol versus adapalene or tretinoin for acne. The Chaudhuri gene-expression data hints at sebum-related pathway activity but I would not switch off a working acne retinoid.
Why do brands keep saying bakuchiol is “as effective as retinol”?
Because Dhaliwal 2019 showed numerical parity at 12 weeks with n=44. That single sentence does heavy commercial work. It is true at the level of that one study. It is not the same as saying the ingredients are clinically interchangeable across populations, durations, and indications.
What concentration should I look for?
Most published work uses 0.5%, occasionally up to 1%. There is no evidence that higher concentrations perform better in finished formulas, and stability of bakuchiol in some bases is its own formulation question.
Related Elelaf tools
Sources
- Dhaliwal S, Rybak I, Ellis SR, et al. Prospective, randomized, double-blind assessment of topical bakuchiol and retinol for facial photoageing. Br J Dermatol. 2019;180(2):289-296. PMID: 29947134
- Chaudhuri RK, Bojanowski K. Bakuchiol: a retinol-like functional compound revealed by gene expression profiling. Int J Cosmet Sci. 2014;36(3):221-30. PMID: 24471735
- Bluemke A, Ring AP, Immeyer J, et al. Multidirectional activity of bakuchiol against cellular mechanisms of facial ageing. Int J Cosmet Sci. 2022;44(3):377-393. PMID: 35514228
- Mukherjee S, Date A, Patravale V, et al. Retinoids in the treatment of skin aging. Clin Interv Aging. 2006;1(4):327-48. PMID: 18046911
- Draelos ZD, Gunt H, Levy SB. Natural Retinol Analogs: A Clinical Evaluation. J Drugs Dermatol. 2020;19(12):1181-1183. PMID: 33346519