Kojic acid: its best evidence comes from being added on top of a gel that already worked

A 1999 trial. A randomised split-face design in forty Chinese women with epidermal melasma.

• Both sides of the face: a gel with 2% hydroquinone + 10% glycolic acid
• One side only, added: kojic acid 2%
• Which side got the kojic acid was randomised
• Assessed every four weeks, ending at 12 weeks

The value of this design is clear. The two sides received something identical except for kojic acid, so any difference belongs to kojic acid. The problem we recorded on the exosome page — 'both sides got the device, so you cannot separate them' — does not arise here.

The results:

• Both sides improved.
• More than half the melasma cleared in 24/40 (60%) on the kojic side and 19/40 (47.5%) on the other.
• Two people cleared completely, and in both cases it was the kojic acid side.

The author's conclusion: adding kojic acid to a hydroquinone and glycolic acid gel improves melasma further.

The conditions to attach when reading this.

First, the base is hydroquinone. In Korea this gel is not a cosmetic. So it cannot be carried over to 'apply a cosmetic containing kojic acid and you get this'. What the trial showed is the increment on top of an already strong treatment.

Second, the abstract gives no P value for this comparison. 60% against 47.5% is a difference in forty people, and the abstract alone does not establish statistical significance. The authors only state that they used Wilcoxon's rank sum test.

Third — and this one favours kojic acid. The side effects (redness, stinging, exfoliation) appeared on both sides of the face and settled by week three. Appearing equally on both sides means that irritation belonged to the glycolic acid and hydroquinone, not to kojic acid.

Fourth, forty people, twelve weeks.

A 2025 randomised, open-label comparative study in India. Seventy-two adult women with facial melasma were randomised 1:1 and applied their cream nightly for 16 weeks.

• Kojic acid 2% cream
• Cysteamine 5% cream

At week 16:

• Mean percentage reduction in mMASI: kojic 10%, cysteamine 12.25%
• Melanin content in lesions by Mexameter: down 4.81% with kojic, 5.13% with cysteamine
• Neither difference was statistically significant.
• No adverse events were reported in either arm.

The authors' conclusion leans toward cysteamine, but we record as it stands that it was ahead on the numbers only, not significantly.

Two limitations here are large.

One, it was open-label. Participants and investigators both knew which cream was which. Melasma assessment runs through human eyes, so this design moves the result.

Two, the absolute values are small. Over 16 weeks, mMASI fell 10 to 12%. In the 1999 trial above, close to half the participants had more than half their melasma clear. Whatever explains that gap, the presence or absence of hydroquinone in the base is likely part of it.

And the comparator is another ingredient. With no untreated side, this trial alone cannot rule out that neither worked. It is the recurring problem in this layer, and the PRP page sets out why it matters in practice.

Hence 'early evidence' for this item.

In Korea, the ingredients accepted as 'functional cosmetic — helps brighten the skin' without efficacy data are listed in Annex 4 of the Regulation on the Review of Functional Cosmetics. There are nine:

Broussonetia extract, arbutin, ethyl ascorbyl ether, oil-soluble licorice extract, ascorbyl glucoside, magnesium ascorbyl phosphate, niacinamide, alpha-bisabolol and ascorbyl tetraisopalmitate.

Kojic acid is not on that list.

Given how long kojic acid has been used as a cosmetic ingredient, that may sound surprising. But the list names ingredients for which efficacy data need not be filed separately — it is not a list of permitted ingredients. An ingredient outside the list can still be used in cosmetics; what it cannot do without individual review is carry a functional claim.

The practical implication for a reader: if a product with kojic acid carries a 'brightening functional' label, that label is usually owed to niacinamide or arbutin also in the formula. The ingredient list will tell you.

On concentration, we found no Korean reference we could verify. The trials above used 2%.