Peroxide whitening softens enamel measurably in the lab, and only in the lab

Peroxide tooth whitening does produce a measurable reduction in the surface hardness of enamel, and the entire evidence base for that finding consists of laboratory studies on extracted teeth. A 2026 systematic review and meta-analysis in the Journal of Dentistry pooled nine such studies and found a pooled ratio of means of 0.89 (95% confidence interval 0.84 to 0.94; p < 0.01) for enamel microhardness after bleaching — a change the authors describe as small, outlier-sensitive, and without clear evidence of clinically meaningful effect [s1]. The complaint with much better clinical evidence behind it is not enamel damage at all. It is tooth sensitivity [s2].

What the enamel meta-analysis did

The review followed PRISMA 2020 and Cochrane Handbook guidance, was registered on PROSPERO (CRD42023472326), and used eligibility criteria aligned with ISO 28399, the international standard for tooth-bleaching products [s1]. It searched four databases for in vitro studies measuring human enamel Vickers or Knoop microhardness before and after hydrogen- or carbamide-peroxide bleaching [s1]. Of 81 eligible studies, nine entered the quantitative synthesis [s1].

The pooled multilevel random-effects analysis, using log-transformed ratios of means, gave ROM = 0.89 (95% CI 0.84 to 0.94; p < 0.01) [s1]. The authors set their own bar for a clinically meaningful change at a ROM exceeding 10%, and concluded that the pooled effect did not clearly meet it — a judgment worth noting sits in tension with a point estimate on the far side of that threshold and a confidence interval spanning both sides of it [s1]. Residual heterogeneity was moderate (tau = 0.074; I² = 50.8%) [s1].

Meta-regression found exposure time to be a significant moderator: a tenfold increase in exposure corresponded to a 6.8% decrease in Knoop hardness (95% CI 13.0% to 0.2%; p = 0.044) [s1]. That association did not survive removing one influential study, after which the estimate fell to a 1.9% decrease with a confidence interval running from a 6.8% decrease to a 3.3% increase [s1]. A finding that flips to null on the removal of a single study is not a finding to plan around. The time-course analyses also pointed in a counterintuitive direction, suggesting larger changes at lower peroxide concentrations [s1].

The review's stated clinical position is cautious in both directions: within the limits of in vitro evidence, peroxide bleaching appears unlikely to cause clinically meaningful harm to enamel integrity when used according to recommended protocols, and the current evidence does not strongly support strict regulatory peroxide limits [s1]. The authors call for more standardised, clinically relevant outcome measures rather than isolated laboratory hardness measurements [s1].

Why "in vitro" is the whole caveat

A microhardness reading taken on an extracted tooth sitting in a dish is not the same measurement as enamel in a living mouth. Saliva remineralises enamel continuously; a tooth in a lab has no saliva, no pellicle refreshing on its surface, and no fluoride from toothpaste arriving twice a day. Every one of the nine pooled studies was in vitro by design [s1]. That does not make the reduction imaginary. It does mean the quantity measured is the response of dead enamel to peroxide with the mouth's repair mechanisms removed, and it is not obvious in which direction that biases the answer.

There is no equivalent body of clinical evidence showing that people who whiten their teeth go on to lose enamel, wear through, or develop more decay. That absence is a genuine gap rather than a reassurance.

What the clinical evidence is actually about

A 2026 umbrella review in the Saudi Dental Journal synthesised 24 systematic reviews of in-office bleaching protocols, searched to January 2026 and appraised with AMSTAR-2 and GRADE [s2]. Its findings are almost entirely about sensitivity, which is the outcome the clinical literature has bothered to measure.

Moderate-certainty evidence indicated that low-to-medium hydrogen peroxide concentrations — approximately 25% to 35% — produce whitening outcomes comparable to higher concentrations while reducing tooth sensitivity [s2]. Light activation, whether LED, halogen or laser, generally did not improve whitening efficacy and may increase sensitivity [s2]. A single gel application produced similar whitening to repeated gel renewals, with a tendency toward lower sensitivity [s2]. Desensitising agents, particularly potassium nitrate and sodium fluoride, consistently reduced sensitivity without affecting whitening efficacy [s2].

One finding in that review deserves flagging for what it says about the industry rather than the biology: the lights are the visible, expensive part of an in-office whitening appointment, and the moderate-certainty evidence says they do not make teeth whiter [s2]. A recent meta-analysis suggesting violet-light activation combined with carbamide peroxide may enhance whitening without increasing sensitivity is noted in the review as resting on limited supporting evidence [s2].

The summary a reader can use

The enamel-damage question, as currently studied, produces a small laboratory signal that the reviewing authors themselves do not treat as clinically meaningful, and no clinical evidence either way [s1]. The sensitivity question has better evidence and a clearer answer: it is common, it is worse at higher peroxide concentrations for no gain in whiteness, and potassium nitrate and sodium fluoride reduce it [s2].

This article describes published research and is not dental advice. Whitening is not appropriate for every tooth or every restoration, and discoloration can have causes that bleaching will not address; those are questions for a dentist.