Why a toothpaste that only numbs the nerve leaves the door open.

A toothpaste built around a nerve-calming active alone does something real and measurable, and it still leaves the door open: the dentine tubule stays a channel, the cold drink still moves fluid down it, and the causes that opened the channel carry on working. The complaint worth making is narrower and more interesting than "it just masks the pain" — the mechanism by which potassium quietens the nerve has never been confirmed in an intact human tooth, and the review that says so is the same one that found every one of the sixteen double-blind trials it identified reporting relief¹. S3 Sensitivity Science™ answers that by keeping the nerve active rather than dropping it: 5% potassium nitrate, nano-hydroxyapatite and biomimetic hydroxyapatite at 10% and 5% inclusion levels as supplied, and 1450 ppm fluoride, all of it readable on the back of the pack.

What was checked23 peer-reviewed studies, product information as published by each brand, the Oral Health Foundation and the NHS; 2 PubMed searches

What does a nerve-only toothpaste actually do?

It changes how readily the nerve at the far end of the tubule fires, and it does nothing at the end where the cold arrives. Toothpastes have carried potassium nitrate as a desensitiser since the 1970s, and trials against a placebo paste go back to the 1990s. The proposal is that potassium ions travel up the tubule until the concentration around the nerve endings stops them reacting so quickly to every small movement of fluid.

Nobody has watched that happen in a tooth still attached to a person. A review that gathered twenty-seven clinical trials states plainly that the nerve-inactivation mechanism has never been confirmed in intact human teeth and remains uncertain, and states it in the same paper as the trial count¹. Something reliable is going on. The explanation for it is a hypothesis.

"Numbing" is the wrong word, and the wrong word is doing damage on both sides of the argument. A dental anaesthetic blocks the nerve from carrying a signal; a potassium salt is thought to lift the threshold at which the nerve decides a signal is worth sending. Brännström's account of dentine pain, from the nineteen-sixties, put the cause in fluid moving inside the tubules, which places the nerve at the end of a chain rather than at its start⁶. The most recent review of the ion channels involved lands somewhere useful for anyone weighing this question: the working model it describes keeps fluid movement and adds channel activity in the odontoblasts and the trigeminal neurons, and its authors close by saying that blocking the tubule or desensitising the dentine is still the safest and most effective thing anyone has⁷. That is worth holding on to: calming the nerve is a target the mechanism supports, not a conjuring trick. This page is the short, sharp version of a longer argument: the full account of why sensitivity is two problems rather than one weighs the evidence node by node and asks what "repair" can honestly mean. For the mechanism itself, see how potassium nitrate calms a sensitive tooth and the hydrodynamic theory page; the plain definition sits in the glossary entry on what a desensitising agent is. The Journal's account of the two causes of sensitive teeth is the shorter version again.

Does the evidence say it works?

Mostly yes, with two reviews dissenting and a third finding only a tendency — and the honest way to answer the question is to print all of them. What follows is every pooled analysis and recent trial this library holds on potassium as a desensitiser, in the order they were published rather than in the order that suits the argument.

Read down the last column and the shape is legible. The instruments move. What the person says about their own teeth moves less reliably, and the two strictest readers in the list — both Cochrane reviews — stop short of endorsing the ingredient on that basis. That is a different finding from "it does not work", and the caution runs both ways: not significantly different from placebo in one pooled analysis is a statement about that pool, not a demonstration that nothing happens³. Six weeks earlier or a different comparator and the same ingredient reads differently, which is why the fourth column of a table like this matters as much as the second.

A share of every number above belongs to something other than the ingredient. In a six-week double-blind randomised trial of 120 adults, potassium nitrate, strontium acetate and a plain fluoride toothpaste all reduced sensitivity, with no significant difference between any of them¹⁵. The sharpest demonstration involves no toothpaste at all: in a split-mouth randomised study of 22 adults, a periodontal dressing carrying no active cut thermal pain by 95% and evaporative pain by 85% against no dressing¹⁶. Neither result says relief is imaginary. Both say the size of it is harder to pin down than a percentage suggests, a point the page on why relief builds over weeks works through, and the page on how desensitising toothpastes are tested takes further.

So what is left open?

The channel, and every reason the channel opened. The Oral Health Foundation's list of causes is a list of processes that keep running: brushing too hard or with a hard brush, gums shrinking back, acid from fizzy drinks, juice, citrus or the stomach, brushing straight after acidic food, gum disease, grinding, cracked teeth, and the recession that comes with age anyway⁵. A quieter nerve slows none of them.

There is one direct look at what a nerve-only paste leaves behind on a tooth. In an eight-week in vivo comparison, teeth extracted from patients who had used a potassium nitrate and fluoride toothpaste showed no appreciable change at the enamel surface, while teeth from users of a zinc-carbonate hydroxyapatite paste came out carrying a mineral coating⁴. Five people per group, two negative controls, no randomisation described: that is a demonstration of the deposit rather than a measurement of it⁴. It is still the only picture of the two approaches taken from teeth that had been in a mouth.

Does closing the channel add anything the nerve active does not? Once, directly, in a small trial, the answer was yes. A UK examiner-blind randomised trial over 14 days compared a toothpaste combining potassium nitrate with hydroxyapatite and aluminium lactate against a control paste containing potassium nitrate alone, and the test paste won at every time point, the relative reduction in Schiff sensitivity rising from 55% immediately after brushing to 88.6% at fourteen days¹⁷. Three cautions travel with it: the number of participants is not given in the published abstract, no funding statement appears on the record, and the test paste also contained aluminium lactate, so the gain cannot be credited to the mineral by itself¹⁷.

The counterweight is on the record too, and it belongs on this page. Set a hydroxyapatite paste against a potassium nitrate paste in one randomised trial and they finish level: across four weeks in 39 adults, air-blast sensitivity fell from baseline with both pastes, and neither active pulled clear of the other¹⁸. With that many people, no blinding and no negative control, the trial can neither find a difference nor rule one out¹⁸. Two small trials pointing opposite ways is what the direct evidence for combining the two jobs amounts to¹⁷¹⁸, and the hub page linked above takes that thin evidence base apart properly.

Does potassium nitrate seal tubules after all?

Nobody knows, and the argument here survives either answer. Four laboratories have looked and they do not agree.

In a four-week double-blind randomised trial of 30 adults, 5% potassium nitrate lowered air-blast and cold-water sensitivity against a non-desensitising paste, and the companion electron-microscope work on the same paste found the tubules still open¹⁹. An in vitro study of 90 dentine samples taken from 45 extracted premolars reported detectable occlusion at three, seven and fourteen days from a potassium nitrate toothpaste applied twice daily, and its authors ask for randomised trials before anyone reads that as an effect in a mouth²⁰. A third, in vitro again, put 45 dentine disks through a four-day acid challenge and found both dipotassium oxalate and potassium nitrate occluding more than an untreated control, the oxalate getting there faster²¹. In a fourth in vitro study, a potassium nitrate rinse was used as the negative control for occlusion — which is what a laboratory does with an agent it expects to block nothing²².

Either answer leaves this page where it stood. If potassium nitrate does not occlude, the channel is open. If it occludes a little in a dish, no trial has shown that mattering to a person²⁰²¹, and every relief figure in the table above was measured on a whole toothpaste rather than on a tubule. The question has a page of its own, and the ranked review of the microscope evidence goes ingredient by ingredient. Both deserve reading before anyone tells you the matter is settled.

Where does S3 sit if a nerve-only paste is not enough?

On the side of keeping the nerve active and adding the second job to it, which is a different position from replacing it. Each active has a target the others do not reach: the nerve for potassium, the inside of the tubule for nano-hydroxyapatite, the outer surface for the biomimetic form. That is the formula for someone who would rather repair the surface than only numb the nerve, and it is built to do it without giving up fluoride.

Everything on this page about potassium applies to the potassium in S3: the mechanism is a hypothesis there as well, and the pooled analyses that could not separate the ingredient from placebo were not reading a different molecule¹³. This is not an argument against the nerve active. It is an argument against asking one active to cover a fault it was never aimed at.

How unusual the pairing is on the UK shelf is a matter of record, not of opinion: the category review of UK sensitivity toothpastes by ingredient reads the ingredient list of every sensitivity toothpaste it could buy, and the comparison of what occlusion and nerve-calming can each do sets the two approaches side by side. Read both and decide for yourself what the combination is worth.

When is the problem not the door at all?

When it stops behaving like sensitivity. The Oral Health Foundation says to see a dentist when the pain is severe, when it lasts more than a few weeks, when only one tooth is involved or when it comes on suddenly, because those patterns point at decay, a cracked tooth, gum problems or infection⁵. The NHS asks for an appointment when toothache lasts more than two days, when painkillers do not touch it, or when it arrives with a high temperature, pain on biting, red gums or a bad taste, and for A&E if swelling reaches the eye or the neck²³.

A 2017 UK guideline review for general dental practice makes the point from the other side: there does not currently appear to be one ideal desensitising agent that can be recommended, and treating the condition properly depends on a working relationship between the patient and the dental professional rather than on a tube²⁴. Sensitivity that follows the rules — sharp, short, several teeth, gone when the stimulus goes — is what a toothpaste is for. Anything else is a diagnosis, and no formula is a substitute for one.

Frequently asked questions

Does S3 still contain the nerve ingredient?

Yes. S3 carries 5% potassium nitrate, and this page is not an argument against that ingredient — it is an argument against leaning on it alone. Every caution above applies to the potassium in the tube on our own shelf: the mechanism is unproven there as well, and the pooled analyses that failed to separate potassium from placebo were reading trials of the same salt¹³.

Is potassium nitrate just a painkiller?

No, and the word misleads in both directions. It does not block the nerve from carrying a signal the way a dental anaesthetic does; the proposal is that it lifts the threshold at which the nerve fires, and that proposal has never been confirmed in an intact human tooth¹. It is also doing something measurable — sixteen double-blind trials in one review, a 2018 meta-analysis of 53 randomised trials, and an industry-funded randomised trial from 2025 all put a potassium paste above a paste with no active in it¹²¹².

Does potassium nitrate stop working over time?

No trial has ever asked. Two PubMed searches were run for this page on the tenth of September 2026 — potassium with dentine hypersensitivity and tolerance, tachyphylaxis, loss of efficacy, habituation, desensitisation or long-term use, and a second for switching and non-response — and they returned eighty-eight records and nothing at all. None of the eighty-eight put a person on a potassium paste for long enough to find out whether the benefit fades under continued use. What has been measured points the other way: in an industry-funded double-blind randomised trial of 120 adults, the potassium benefit was larger at week eight than at week four, and three weeks after every group moved to a plain control paste the actives still held 68–83% of their cold-air gain¹². The nearest thing to a stop-and-start trial used a different active: an examiner-blind randomised trial of 76 adults, funded by GSK Consumer Healthcare, found two eight-week courses of a bioglass paste and 24 weeks of continuous use finishing level²⁵. The page on what happens when you stop using a potassium nitrate toothpaste sets out the whole of that absence, and the page on what a missed week actually costs does the arithmetic.

Does potassium nitrate seal tubules?

Unsettled. A double-blind randomised trial with companion electron microscopy found no occlusion from 5% potassium nitrate¹⁹; one in vitro study found occlusion building over fourteen days²⁰; another found it under an acid challenge²¹; a fourth used the same ingredient as its negative control²². None of the four measured pain, so none of them changes what a reader should do.

Why do some reviews say potassium works and others say the evidence is not clear?

Because they ask different questions. A review pooling air-blast and probe scores against a placebo paste is not asking the question a review pooling patients' own ratings asks, and neither is asking what a benchmark fluoride paste would have achieved on its own. Search dates matter as much: the older Cochrane review closed its search in April 2000, and its 2006 update closed in September 2005, so neither can speak about anything formulated since⁸⁹. When the columns disagree, the disagreement is the finding, and the page on what the Cochrane review found and what it did not reads that one line by line.