Teeth hurt when breathing cold air: why it happens and which toothpaste ingredients help.

Teeth hurt when you breathe cold air because moving air cools and dries an exposed patch of dentine faster than the tooth can equalise, and the fluid inside its open tubules shifts far enough for the nerve to fire. Which toothpaste helps is, unusually, a question the evidence answers head-on — a puff of cold air is the stimulus dentine sensitivity trials actually measure — and the biggest pooled analysis of it, a 2026 systematic review and network meta-analysis of 93 randomised trials in 9,548 participants, names stannous fluoride and arginine as the first-line self-care options¹. S3 Sensitivity Science™ carries 5% potassium nitrate with 10% nano-hydroxyapatite and 5% biomimetic hydroxyapatite, both as solution, plus full adult-strength fluoride; it contains neither stannous fluoride nor arginine.

What was checked28 peer-reviewed studies, S3 consumer trial (ADSL, 2026), the Oral Health Foundation and the NHS

Which toothpaste helps when cold air hurts your teeth?

Two families of active have a measured effect on a cold-air score, and they work at opposite ends of the same problem: one narrows the open tubule so the air moves less fluid, the other makes the nerve at the far end slower to fire. The narrowing actives with cold-air numbers are stannous fluoride, nano-hydroxyapatite, arginine and the bioactive glasses; the nerve-calming one is potassium, usually as potassium nitrate¹.

The uncomfortable part, for anyone selling one of these, is which two the reviewers put in front. The authors of the 2026 network meta-analysis conclude that stannous fluoride and arginine toothpastes should be the first-line self-care options, chosen on preference, tolerability and availability rather than on any expectation of superior efficacy¹. S3 contains neither. What this page does with that comes further down, on the evidence rather than on an adjective.

Why does a breath of cold air hurt when a cold room does not?

Because a cold room does not move air across the tooth, and moving air does two things at once: it cools the surface and it dries it. Dentine is not solid. It is threaded with tubules, fluid-filled channels a few micrometres across running from the surface in towards the pulp, and a patch of it only becomes sensitive when the surface is first worn or exposed and the tubules there are then opened to the mouth at one end while still open to the pulp at the other⁸. Evaporation from an open tubule drags fluid outwards, and the nerve endings at the inner end read the movement rather than the temperature. That is the hydrodynamic theory, Brännström's, and every active named on this page is an attempt to interrupt it at one end or the other. The Journal's page on why cold hurts your teeth answers the same question in a paragraph; this is the version with the trial numbers in it.

The shape of the pain has an explanation too, though a much older and thinner one. Single-unit recordings from intradental nerve fibres in anaesthetised cats — animal work, and from 1987 — separated two populations: fast A-delta fibres conducting at a mean of 13.9 metres per second, which fired a burst only while the tooth temperature was actually falling and stopped the moment it settled, and slow C fibres at 1.3 metres per second, which did not begin until a mean of 7.3 seconds had passed⁹. Nothing in that says anything about a person with sensitive teeth. It is, though, the reason the sharp part of a cold breath tracks how fast the tooth is cooling rather than how cold the air is, and the reason it is over before the breath is.

How the cold is sensed at the cellular level is not settled. One line of animal work puts a channel called TRPC5 in the odontoblast, the cell that lines the inner end of the tubule, and makes that cell part of the tooth's cold-sensing apparatus¹⁰; another found that deleting TRPM8 or blocking TRPA1, the two channels anyone would expect to be responsible, did not stop a tooth responding to noxious cold at all¹¹. The page on cold drinks and ice cream takes that argument further. What belongs here is the honest position, which a 2025 review states plainly: the working model is now fluid movement plus ion channels rather than fluid movement alone, those channels do too many other jobs in the body to be drug targets, and occluding the tubule or desensitising the dentine remains the safest and most effective thing available¹².

Which teeth notice first is where the surveys disagree, and the disagreement is worth having. In a 2026 cross-sectional clinical study of 3,551 adults across seven European countries, funded by Haleon, 29.1% had at least one site scoring two or three on the cold-air scale, and sensitivity was commonest on the cheek-facing surfaces and in the lower incisors, the teeth a cold breath reaches first¹³. UK general practice tells a different story. In one cross-sectional clinical study in the South West, twelve dentists examined 3,593 patients over a calendar month and diagnosed dentine hypersensitivity in 137 of them, a prevalence of 3.8%¹⁴. Two years later, in a second cross-sectional clinical study, nineteen practices across the country examined 4,841 patients and diagnosed 201, or 4.1%¹⁵. In both, the upper premolars were the teeth most often affected, and the trigger patients named most often was cold drinks, not cold air¹⁴¹⁵. The gap is less a disagreement about teeth than about counting: one examined everybody with a standard air stimulus, the others counted whoever a dentist happened to diagnose in an ordinary month of appointments.

A cold breath is the test: what an air-blast trial actually does

A dental syringe, one second of air, and a number between zero and three. That is the whole instrument. The consensus guidelines that still shape how sensitivity trials are designed ask for three stimuli — touch, cold and evaporative air — a blinded, parallel-group design with random allocation, eight weeks for most trials, and at least two independent trials before a product is approved². The evaporative one is the stimulus you reproduce every time you step outside in January: in a 2025 industry-funded trial that measured onset unusually carefully, the method reads simply that a one-second application of cold air was delivered onto each tooth from a standard dental unit syringe, and scored on the Schiff index, where zero is no reaction and three is a reaction with a request that the stimulus stop³.

Two things about that number are worth knowing before reading any of the results below. The first is that it is the field's default rather than a validated instrument: a 2026 scoping review of 71 studies found the visual analogue scale and the Schiff cold air sensitivity scale to be the two most used measures, frequently together, and the references authors gave for choosing them inconsistent¹⁶. The second is that the scales have been put through an accuracy test exactly once. In a cross-sectional clinical study of 72 adults, five scales were compared against the participants' own sound teeth, and the study found the Schiff scale the most specific of the five at 91% and the least accurate by area under the curve at 0.729, against 0.750 for a plain numeric rating scale; all five were judged accurate enough to diagnose the condition⁷. The caveat belongs here rather than in a footnote: the stimuli in that work were a probe and an ice stick, so what was tested was the scale, not the air blast⁷.

What a person's own cold-air threshold looks like has been measured properly once as well, and the interesting number is not the average. In a non-randomised clinical study, twenty-nine adults with a single sensitive tooth were tested repeatedly over three weeks with a purpose-built cold-air device: the temperature that evoked moderate-to-strong pain was highly stable within a person, with a single-measure intra-class correlation of 0.83, and averaged −13.69 °C, while the spread between people had a standard deviation of about ten degrees¹⁷. Sensitivity is not one threshold everybody shares and some people cross more often. It is a different threshold per mouth, held steady, and yours is the only one deciding whether a given morning hurts.

That device deserves its own sentence, because it is the only published apparatus that says what the air actually is. In a clinical study describing its construction, it delivered repeatable graded puffs from room temperature, 24.5 °C give or take two degrees, down to −35 °C, at flow rates between five and seventeen litres a minute, and its authors report that the puffs felt like natural cold stimuli¹⁸. The ordinary surgery does none of that. A dental unit syringe blows compressed air at whatever temperature the room happens to be, which is why that group had to build the apparatus at all, and it is one more reason to read a Schiff difference as a comparison of two toothpastes, not as a physical measurement of your morning.

One last piece of the stimulus has been measured: how long a tooth needs before it will give the same answer again. Forty adults per stimulus group were re-stimulated at ten minutes, five, two, and immediately after the first pain had stopped, in a non-randomised clinical study; with a cold air blast the two-minute response was much like the ten-minute one, while a probe drawn across the dentine needed longer to recover¹⁹. That is why trial protocols wait between teeth.

And here is what nobody has measured. Five PubMed searches run for this page on the tenth of September 2026 returned no published work on dentine sensitivity to outdoor air, only to the clinic's syringe. Every figure below was produced indoors, by a device, on a tooth someone was holding still. The step from there to a bus stop in January is an inference, and this page makes it in the open rather than quietly.

One questionnaire survey of dental-clinic patients with sensitive teeth — not a trial, and with no cold air anywhere in it — put self-reported mouth breathing at 16.4%²⁰. The page on exercising outdoors takes the deliberate version, where breathing through the mouth is not a choice.

What does the pooled evidence say, active by active?

It says more about cold air than about any other everyday trigger, because the number written down in the clinic is the cold-air score itself. Here is the whole of it, and then the parts that disagree.

The anchor is that 2026 systematic review and network meta-analysis in Periodontology 2000, which pooled 93 randomised trials in 9,548 participants and ran its network comparisons on 47 trials and 4,636 participants, grading the certainty of every estimate¹. Measured at two weeks against a benchmark fluoride toothpaste, the meta-analysis put the mean differences on the evaporative cold score at: nano-hydroxyapatite −0.96 (95% CI −1.40 to −0.52) from two studies at moderate confidence; stannous fluoride −0.85 (−1.08 to −0.62) from ten studies at high confidence; arginine −0.78 (−1.02 to −0.54) from seven studies at low confidence; potassium with stannous fluoride −0.75 (−1.54 to 0.04) from one study at low confidence; strontium acetate −0.52 from one study at low confidence; potassium with or without fluoride −0.42 (−0.77 to −0.07) from one study at low confidence, which the reviewers describe as a small but important effect; bioglass −0.36 (−0.74 to 0.01) from three studies at low confidence; and placebo −0.13 from two studies¹.

Four things in that list matter more than its order. The first is confidence: the largest point estimate belongs to nano-hydroxyapatite and rests on two trials, while the confident estimate belongs to stannous fluoride and rests on ten. The second is what the authors did with it, which was to recommend stannous fluoride and arginine as first-line self-care, explicitly not because either beats the rest¹. The third is who paid: the review records that most of the trials it included were industry-funded, at 96% of the stannous studies, 86% of the arginine studies and 76% of the potassium studies¹. The fourth is what is absent. The pairing S3 uses, potassium nitrate together with hydroxyapatite, is not one of the nodes in that network, so the review has nothing to say about it either way.

Three other network analyses have asked the same question and answered it differently, and averaging them would be dishonest. A 2020 network meta-analysis of 125 randomised trials in 12,541 patients ranked the actives separately for each stimulus, and for the air stimulus it gave standardised mean differences of 2.44 for potassium combined with hydroxyapatite, 2.28 for potassium with stannous fluoride, 2.22 for arginine, 1.98 for calcium sodium phosphosilicate and 1.90 for stannous fluoride²¹. In a 2019 network meta-analysis of thirty randomised trials, nano-hydroxyapatite had the highest probability of being the most effective at two and four weeks, at 60% and 67%, and arginine at eight weeks, while potassium was not significantly different from placebo — and a significant placebo effect was measured in the same network⁴. A 2017 network meta-analysis of nine randomised trials found strontium acetate and arginine significantly ahead of placebo, and potassium nitrate a tendency towards relief that never reached significance⁵. A 2018 systematic review and meta-analysis of 53 randomised trials in 4,796 patients placed nano-hydroxyapatite in the moderate-quality GRADE band and potassium-only toothpaste in the low-quality one²².

For potassium the older evidence runs the other way and then qualifies itself. The 2006 Cochrane systematic review pooled six randomised trials and found potassium nitrate toothpastes reduced air-blast sensitivity at six to eight weeks, with a standardised mean difference of −1.25 (95% CI −1.65 to −0.851)⁶. In the same review the patients' own overall assessment did not move significantly, and the reviewers declined to call the case settled⁶. For hydroxyapatite, a 2019 systematic review and meta-analysis of six four-week randomised trials found nano-hydroxyapatite ahead of its comparators on the evaporative stimulus, at a standardised mean difference of −1.09, and on touch, with no difference at all on cold liquids — worth knowing if the cold tap troubles you as much as the walk to the door²³. And the strongest single-active case is the manufacturer's own: an industry-funded meta-analysis of 14 randomised trials in 1,287 participants, taken from Procter & Gamble's archive and written with company scientists among its authors, reported a 57% benefit for stannous fluoride over plain fluoride pastes on the cold-air score, and a 22% advantage over potassium nitrate or arginine positive controls²⁴.

Read across the table below rather than down any one column. The pooled numbers are not interchangeable: they come from different networks, different reference pastes and different years, and where two reviews disagree the cell says so.

How long does it take, and what happens if you stop?

Not days. The systematic review that grouped treatments by when their effect first reached significance, across 74 randomised trials in 5,366 patients, found only in-office procedures — glutaraldehyde with HEMA, glass ionomer cements, laser — in the immediate band of up to seven days; stannous fluoride and hydroxyapatite reached significance in the medium term, up to a month, and potassium nitrate, arginine and hydroxyapatite in the long term²⁵. No toothpaste is on the immediate list. That is the honest calendar, and it is why relief from a tube is judged at two to four weeks rather than on the morning it was bought.

The most detailed onset curve comes from an industry-funded double-blind randomised trial of 120 adults, paid for by a stannous fluoride manufacturer, which scored cold air at day three and at weeks two, four and eight³. In that same industry-funded trial, the week-eight improvements over a plain fluoride control were 57% for stannous fluoride, 47% for an experimental oxalate paste and 44% for potassium nitrate, and the differences between the three actives at week eight were not significant³. The same manufacturer-funded trial then moved everyone onto the control paste for three weeks, and all three groups kept between 68% and 83% of the cold-air gain they had reached at week eight³. An examiner-blind randomised trial of 215 adults, funded by the manufacturer of the paste it tested, had a bioactive glass significantly ahead of a fluoride paste from day three and still improving at week eight²⁶. And an industry-funded twelve-week double-blind randomised trial of 129 adults testing a hydroxyapatite and potassium citrate paste found its air-stimulus improvement still present four weeks after the paste was stopped²⁷.

Two things follow for a tube in a bathroom. Give it eight weeks before deciding, because that is the duration the consensus guidelines ask trials to run². And keep going afterwards: the Oral Health Foundation's own line is that a sensitive toothpaste maintains its effect only while you keep using it²⁸. The Journal's page on how sensitive toothpastes work covers why the wait is a wait, and the resources page on how long to give a new paste before switching sets out the test in full.

Where does S3 sit against a cold breath?

The two jobs are the ones the standard review chapter names: alter the fluid flow in the tubules by occluding them, or modify or chemically block the pulpal nerve⁸. Hydroxyapatite does the first, potassium the second, and neither does the other's job. S3 was formulated to carry both in one tube: potassium nitrate for the nerve, nano-hydroxyapatite inside the tubule, biomimetic hydroxyapatite on the surface, each doing something the others cannot.

The declared levels come with a caveat S3 publishes itself: 10% nano-hydroxyapatite and 5% biomimetic hydroxyapatite are inclusion levels of the ingredient as supplied, and the active hydroxyapatite content is lower. In S3's own consumer trial, run by ADSL and never published in a journal, 84% said their teeth felt less sensitive in trigger situations after one week. That is perception wording from an uncontrolled commissioned panel, and it belongs in a different column from a pooled cold-air difference; this page will not pretend otherwise. Nor will it hide that the Oral Health Foundation's own advice names potassium citrate, potassium nitrate and stannous fluoride as the actives to look for, and does not mention hydroxyapatite at all²⁸.

What S3 cannot claim here is the top of that table: stannous fluoride's estimate is the confident one, on ten studies at high certainty; nano-hydroxyapatite's is the large one, on two studies at moderate certainty; and the reviewers' first-line recommendation is stannous fluoride and arginine¹. S3 contains neither, and says so. What it does claim is narrower and checkable: both of the mechanisms a cold breath calls on are in the tube, at levels printed on the pack, with full adult-strength fluoride kept in, and the effect builds over weeks rather than arriving on the morning the cap comes off.

When is cold-air pain a dentist's question?

When it stops behaving like sensitivity. Dentine sensitivity is short, sharp and tied to the stimulus: it arrives with the breath and leaves with it, which is the pattern the fast fibres would produce⁸⁹. Four departures from that are on the Oral Health Foundation's list of reasons to book: pain that is severe, sensitivity that has lasted more than a few weeks, one tooth rather than several, and a start that was sudden — any of which can point to decay, a crack, a gum problem or an infection²⁸. Two more are worth adding from the timing itself. An ache that carries on after the cold has gone, instead of switching off with it, is not the hydrodynamic pattern. Pain on biting is on the NHS's own list, alongside toothache lasting more than two days, toothache that painkillers do not touch, a high temperature or a swollen cheek — and the NHS says plainly that this is a dentist's job and not a GP's²⁹. The page on being sensitive to hot and cold at once sets out the timing test in more detail.

Frequently asked questions

Which toothpaste ingredients in S3 have cold-air evidence?

Two of them, at different strengths of evidence. Nano-hydroxyapatite has a direct cold-air result: in a double-blind randomised trial of 105 adults, a 15% paste lowered cold-air sensitivity more than a fluoride paste and a placebo at two weeks and again at four³⁰. It also has a pooled two-week estimate at moderate confidence in the 2026 systematic review and network meta-analysis, resting on two studies¹. Potassium nitrate has the Cochrane air-blast result at six to eight weeks⁶ and, in that same network, a small effect from a single study at low confidence¹.

The combination itself has not been trialled as a combination, and no page should imply that it has.

Why do my front teeth notice a cold breath first?

Partly because they meet the air first, and partly because of where the exposed surfaces tend to be. In the cross-sectional clinical study of 3,551 European adults funded by Haleon, sensitivity was commonest on the cheek-facing surfaces and in the lower incisors¹³. Two UK practice surveys pointed instead at the upper premolars and named cold drinks as the trigger patients reported most¹⁴¹⁵. Both can be true at once: the first examined every participant with a standard air stimulus, the second counted only the people a dentist diagnosed.

How long should I give a sensitivity toothpaste before judging it on cold air?

Eight weeks, which is what the consensus guidelines ask of trials, with a first honest look at two to four². Nothing sold in a tube appears in the immediate band of the review that grouped treatments by onset; only in-office procedures did²⁵. If a cold morning still catches you at week two, that is not yet evidence the paste is wrong for you.

Does breathing through my nose or wearing a scarf help?

Yes, in the obvious way and not at all in the way that lasts. A scarf or a nose-breath warms and moistens the air before it reaches the tooth, so the trigger is removed while it is there; the open tubule underneath is unchanged, and it is waiting when the scarf comes off. One questionnaire survey of dental-clinic patients with sensitive teeth — not a trial — put self-reported mouth breathing at 16.4%, with nothing in it tested against cold air²⁰. Treat the scarf as comfort and the toothpaste as the part that works on the tooth.

Is cold-air pain ever a warning of something worse?

Sometimes, and the tell is the timing rather than the intensity. Sensitivity is sharp and finished with the breath. An ache that continues once you are indoors, pain in one tooth rather than several, a start that was sudden, pain on biting, or sensitivity that has run on for more than a few weeks are the ones to take to a dentist rather than to a tube²⁸²⁹.