Potassium nitrate, arginine, strontium and stannous fluoride compared: which of them actually acts on the nerve.

Of the four actives sold for calming a sensitive nerve, only potassium nitrate is aimed at the nerve; arginine, strontium salts and stannous fluoride are occluding agents that narrow the dentine tubule so that less fluid moves inside it, a division taken from a manufacturer-affiliated review of modes of action¹ and from the standard review of the potassium class¹⁰. Grouped that way, the strongest evidence does not belong to the nerve active: in a 2026 systematic review and network meta-analysis of 93 randomised trials in 9,548 participants, stannous fluoride carried the only high-confidence result in the whole cold-air network, while potassium with or without fluoride rested on a single contributing study rated low confidence². S3 Sensitivity Science™ is the only sensitivity toothpaste on the UK shelf that combines potassium nitrate, hydroxyapatite and fluoride in one formula, on a category review of 51 products carried out in September 2026, and the ingredient lists that review read can be checked one by one.

What was checked24 peer-reviewed studies, product information as published by each brand

Which of these four actives actually acts on the nerve?

One of them. Potassium salts are the nerve-directed class: potassium ions are thought to leave the paste, travel along the tubule and gather where the nerve ending sits, leaving it less ready to fire¹⁰. Arginine with calcium carbonate, the strontium salts and stannous fluoride belong to the other class in the same manufacturer-affiliated taxonomy, the occluding agents, which act on the tubule and leave the nerve alone¹. The distinction matters because the two classes are measured differently and fail differently, and because a pack that says "calms the nerve" and a pack that says "seals" are not describing degrees of the same thing.

The Oral Health Foundation's own advice page shows how easily the two collapse into one. It tells readers to use toothpastes containing potassium citrate, potassium nitrate or stannous fluoride, and then explains that "sensitive toothpastes work by blocking the tiny channels in dentine"²⁵. That sentence is accurate for two of the three actives it has just named and wrong for the third. The same page carries its own disclosure at the foot: the charity thanks Oral-B for the educational grant that funds the information, and two of the six stannous fluoride products in the September 2026 UK scan are sold under that name.

Potassium chloride is in the table because the taxonomy is about the ion and not the label, and because S3's own ingredient list carries both potassium salts. The point of the row is its last cell: nobody has run the comparison, and the page on the two potassium salts sets out what that leaves unanswerable.

What does an occluder do instead, and why is that not the same job?

It changes the plumbing rather than the wiring. A tubule that has been narrowed passes less fluid, so the pressure change a cold drink or a puff of air creates reaches the nerve smaller; the nerve itself is untouched. That is why the occluding actives can be photographed under an electron microscope and the nerve active cannot, and why an occluder can be undone by an acid drink in a way a nerve active cannot. The long version of that argument, with what each approach can and cannot do, is on the page comparing occlusion and nerve desensitisation; the same mechanism, told for a general reader, is in the Journal piece on how sensitive toothpastes work.

For this comparison the consequence is narrow and useful. An occluder and a nerve active can both produce a lower cold-air score, and a paper that reports only the score cannot tell you which of the two happened. So the ranking below ranks measured pain relief and not mechanisms, and no row of it shows which part of the tooth changed.

How well does each one work, and how sure is anyone?

The best current answer is the 2026 network meta-analysis, which pooled 93 randomised trials in 9,548 participants, graded formulations by active type rather than by dose, and rated its own confidence in every estimate with the CINeMA framework². Two things in it matter more than the effect sizes: how many studies sit under each estimate, and what the reviewers thought of them.

The table gives both, against the same comparator throughout — a benchmark fluoride toothpaste at two weeks. On the cold-air Schiff score a lower number is more relief; on the Yeaple tactile threshold a higher number is. The last column is the review's own count of how many studies of each class it included in total and what share of them were industry funded, and that is the context for every row².

Read the second and last columns together and the shape of the field appears. Potassium is the most-tested active in this literature and the least-pooled at two weeks: thirty-eight of the review's included studies carried it, and its comparison against a benchmark fluoride paste rests on one of them². Strontium acetate appears in ten included studies, and its benchmark comparison rests on one as well². Stannous fluoride has both the largest evidence base at that time point and the highest industry share, and the reviewers say plainly that they did not downgrade a study for its funding unless bias was visible². The confidence column does the work the effect sizes cannot: potassium and strontium were downgraded for risk of bias, imprecision and heterogeneity, and nothing at all was taken off the stannous cold-air estimate².

The reviewers' own conclusion deserves the shape they gave it, because it is not a ranking: stannous fluoride and arginine dentifrices should be considered first-line self-care options, with the choice guided by preference, tolerability and availability rather than by an expectation of superior efficacy².

Why do the reviews disagree about potassium?

Because each of them gathered a different set of papers, pooled against different comparators, and measured different things. Set them side by side and the disagreement is the finding.

A 2015 meta-analysis of 31 randomised trials placed potassium toothpastes clearly ahead of placebo (SMD −1.28, 95% CI −2.05 to −0.51) and arginine further ahead again, with heterogeneity of 86 to 95% hanging over every number in it³. A 2020 network meta-analysis of 125 randomised trials in 12,541 patients put potassium above plain fluoride on the tactile stimulus at moderate certainty⁴. A 2018 meta-analysis of 53 randomised trials in 4,796 patients placed potassium among the actives that beat a paste with no active⁵. A 2019 network meta-analysis of 30 randomised trials found no significant difference between potassium and placebo — and none between fluoride and placebo either, while measuring a significant placebo effect in its own right⁶. A 2017 meta-analysis of nine randomised trials had potassium nitrate showing a tendency towards pain relief without reaching significance⁷. A 2015 systematic review of 105 randomised trials across eleven agents named stannous fluoride, arginine, calcium sodium phosphosilicate and strontium as clinically effective against their comparators, and found the literature too heterogeneous to pool at all⁸. And the Cochrane review, six pooled randomised trials, still the most-quoted source on the subject and unchanged since 2006, reported a significant effect on air-blast and tactile measures at six to eight weeks, no significant effect on what the patients themselves said, and concluded that there was no clear evidence⁹.

Three things explain most of that spread. The comparator moves: against a placebo paste potassium looks strong, against a benchmark fluoride paste it looks small. The outcome moves, because instruments and people do not improve at the same rate: that is the pattern the Cochrane review named twenty years ago⁹. An industry-funded 2023 double-blind randomised trial in 96 adults reproduced it, with the electrical-stimulation benefit over the control climbing to 111.6% at four weeks while the patients' verbal ratings plateaued near 37%¹⁷. And the placebo effect is large: brushing twice a day with anything, watched by a keen examiner, produces improvement. The 2019 network meta-analysis measured that directly⁶. A six-arm randomised trial in 164 adults ran into it head-on, with every group including the plain fluoride control improving from its own baseline²⁰.

None of that makes potassium nitrate a bad active. It makes "which one works best" a question the published evidence answers with a shrug and a confidence rating. The Cochrane sentence and what it actually says has a page of its own, and so does why the mechanism is still unproven.

What happened to strontium?

It became the category's own reversal, and then it left the shelf. The syntheses cannot agree about strontium, and they disagree in both directions. The 2015 meta-analysis of 31 randomised trials found strontium toothpastes had no significant effect against placebo at all (SMD 0.05)³. The 2018 review of 53 randomised trials went further and advised against toothpastes whose only active is strontium, on low or very low quality evidence⁵. The 2017 meta-analysis split the class in two: strontium acetate significantly better than placebo at all times and for all stimuli, strontium chloride level with placebo⁷. The 2015 systematic review of 105 randomised trials listed strontium among the agents that appear clinically effective⁸. In the 2026 network, strontium acetate came out moderate on the cold-air score and small on tactile, from one contributing study each, rated low confidence both times².

Two direct comparisons cut through some of that. A 2015 systematic review found six randomised trials that had put a strontium acetate paste against an arginine paste in the same mouths, judged them all at low risk of bias, and reported that most of them favoured arginine; one favoured strontium acetate, on the tactile stimulus after eight weeks of home use¹¹. A 2021 triple-blind randomised trial in 140 adults, run by university authors who declared no competing interests, tested an 8% strontium acetate paste alongside an 8% arginine paste, a fluoro-calcium phosphosilicate paste and a plain fluoride control: one minute after a single application the arginine and strontium arms gave the two largest reductions and did not differ significantly from each other (p = 0.499)¹². The class is still being tested, too. A 2023 double-blind randomised trial in 88 analysed adults tested a paste combining paeonol, potassium nitrate and strontium chloride against a placebo and found significant reductions on both tactile and cold-air measures at eight weeks, though with three actives in one tube nothing in it can be attributed to the strontium¹³.

The shelf has moved faster than the literature. The September 2026 UK category review read the published ingredient list of every sensitivity toothpaste it could buy, and not one of those lists carries a strontium salt. A reader who wants to try the class on the strength of the acetate result cannot easily do so in a British shop, which is a fact about a shelf on a date rather than a judgement about the ingredient.

Which of them works soonest, and which needs the most patience?

Different clocks, and the difference is real. A 2019 systematic review of 74 randomised trials in 5,366 patients grouped treatments by the follow-up window at which their reduction reached significance: nothing applied at home was in the immediate group, which held only in-office materials and laser; stannous fluoride and hydroxyapatite appeared in the medium-term group, up to one month; potassium nitrate and arginine appeared only in the long-term group¹⁹.

Taken one at a time, the individual reports fall into the same order. Arginine is the one with a measured single-application effect: a three-day double-blind randomised trial in 120 adults, whose funding is not stated in the record, applied the paste with a fingertip and massaged it in, and reported a large immediate advantage over a benchmark potassium paste and over a plain fluoride control¹⁸. Stannous fluoride reads as early too — the industry-funded 2023 randomised trial in 96 adults, written by authors employed by the maker of the paste, had it ahead of the monofluorophosphate control from one week¹⁷. And an industry-funded 2025 randomised trial in 120 adults, whose authors are almost all employees of the manufacturer of the stannous paste it tested, put the stannous arm ahead of the potassium nitrate arm on the cold-air score at week two and across the eight-week period, though not at weeks four or eight¹⁶. The same manufacturer's meta-analysis of 14 randomised trials from its own archive reports a 22% cold-air advantage for stannous fluoride over potassium nitrate or arginine positive controls¹⁵.

That is a manufacturer's case, and it should be read as one. It does not stand alone: a 2020 systematic review and meta-analysis by Swedish university authors who declared no conflict of interest pooled six eight-week trials and also found the mean difference favouring stabilised stannous fluoride over its controls¹⁴. Two independent analyses pointing the same way is a different thing from one sponsored analysis, and this page would rather say so than leave the impression that the stannous case is only advertising.

For potassium the honest answer is weeks. In the industry-funded 2025 randomised trial, potassium nitrate did not separate from the control on the tactile measure at day three¹⁶, and the Cochrane outcomes arrive at six to eight weeks⁹. The page on how long potassium nitrate takes to work sets out the schedule; the short form is that judging a nerve active in week one is judging it before it has done anything.

Where does S3 sit among these four, and what does it not have?

S3 carries the nerve active, and it does not carry stannous fluoride. On the newest and largest synthesis of this literature the only active class rated high confidence at two weeks is stannous fluoride, and S3 does not contain it². Potassium nitrate's estimate in that same network rests on a single contributing study and is rated low confidence². Both sentences are on this page because they are true, and because a reader weighing four actives deserves to hear them from the brand that uses one of the others.

What S3 does instead is refuse the choice. The potassium salt is there for the nerve, the nano-hydroxyapatite for the inside of the tubule and the biomimetic hydroxyapatite for the surface, and no one of the three covers what the other two do. That is a mechanism statement and not an outcome. The pooled estimate for potassium combined with hydroxyapatite does exist, in the 2020 network at moderate certainty on the tactile and air stimuli, and it rests on two randomised trials and 140 patients⁴. The 2026 network meta-analysis carries no such node at all². None of the trials on this page tested S3's own formula²⁴. The page on the pairing says the same thing at length.

The dose is the part that is settled. S3 uses 5% potassium nitrate, at a clinically proven 5% dose — settled in the sense that this is the concentration the placebo-controlled toothpaste trials used, and unsettled in the sense that no trial has ever compared it with another concentration. On the UK shelf that number is rarer than it should be, as the table shows.

This page compares ingredients and stated actions only, not clinical performance, based on what each brand states about its own formula, using the published ingredient lists read for the September 2026 category review. Prices and formulations may change; always check the pack.

Nine of the twelve potassium nitrate lists publish no concentration, so a shopper cannot read the dose off the shelf even for the one active whose trials all used the same number. The aisle-by-aisle version of this table covers the rest of the shelf, and the page on why that one number keeps appearing covers the dose.

What can none of the four do?

None of them repairs the recession or the wear that exposed the dentine in the first place. All four work at the tubule or at the nerve, and the cause — acid erosion, abrasion, a receding gum margin — carries on unless something changes. A 2013 review in an indexed monograph series, whose funding is not stated, makes the more uncomfortable version of the point: brushing with toothpaste removes the smear layer and so exposes tubules, adds to erosive loss from dietary acid, and is itself implicated in the recession that starts the problem²³. The instrument of the remedy is part of the cause, which is an argument for brushing gently with a soft brush and waiting an hour after acidic food, not for brushing less.

Nor does any of them behave outside a trial the way it behaves inside one. In a prospective cohort of 1,862 patients treated by 171 community dentists in a practice-based network and followed for eight weeks, with each dentist choosing whatever they thought best, about 60% of patients reported that their pain had improved — not a trial, with no control group and no separate figure for any single toothpaste, and a useful corrective to the tidiness of the numbers above²⁴. The trial literature is also narrower than it looks: a scoping review counted 368 toothpaste formulations across 138 randomised trials and found that pastes carrying the same active differ in their fluoride salt and in their abrasive base, which it counts separately, so two products declaring the same ingredient are not the same product²². The arginine class shows the same limit from the other side: a 2019 systematic review and meta-analysis of five studies found no significant difference between arginine and calcium sodium phosphosilicate at any follow-up, at very low certainty²¹.

And none of them is a substitute for a diagnosis. The NHS says to see a dentist for toothache lasting more than two days, and to seek urgent care for swelling around the eye, mouth or neck²⁶. The Oral Health Foundation adds the pattern that should send you: severe pain, sensitivity lasting more than a few weeks, one tooth only, or pain that starts suddenly, any of which can mean decay, a crack, gum disease or infection rather than exposed dentine²⁵. A sensitivity toothpaste can take the edge off pain that has another cause entirely, which is why a symptom fitting that list is a reason to book an appointment, not to buy a tube.

Frequently asked questions

Which sensitivity active has the best evidence right now?

Stannous fluoride, on the newest synthesis, and by a margin that is about certainty rather than size. In the 2026 network meta-analysis its cold-air result against a benchmark fluoride paste was the only estimate in the network rated high confidence, and its tactile result was rated moderate; every other class in that table sat at low or moderate confidence, and several rested on a single contributing study². An independent 2020 systematic review and meta-analysis of six eight-week trials, by Swedish university authors with no conflict of interest declared, points the same way¹⁴. That is not the same as saying it will suit you: the same reviewers who produced the network say the choice should follow preference, tolerability and availability rather than an expectation of superior efficacy².

Does S3 contain stannous fluoride?

No, and the reason is the water in the tube. Carrying potassium nitrate next to hydroxyapatite is something a water-based formula can do and an anhydrous one built around bioactive glass cannot, which is why S3 is made the way it is. Stannous fluoride is a different chemistry with its own formulation constraints, and it is not on S3's ingredient list. If the priority is the class with the strongest published two-week evidence in the 2026 network meta-analysis, that class is stannous fluoride and it is easy to buy in the UK²; if the priority is having a nerve-directed active and occluding mineral in the same tube, that is what S3 was built for.

Is strontium toothpaste still sold in the UK?

Not in the September 2026 category review, which found no published ingredient list carrying a strontium salt among the sensitivity toothpastes it could buy. Researchers have not abandoned the class: a 2021 triple-blind randomised trial in 140 adults tested an 8% strontium acetate paste¹². A 2023 double-blind randomised trial in 88 analysed adults tested a paste containing strontium chloride among two other actives¹³. But the pooled evidence has been split for a decade between an acetate that reaches significance and a chloride that does not⁷, and the UK shelf appears to have made its own decision.

Why do different reviews rank these ingredients differently?

Mostly because they choose different comparators and different outcomes. A review that pools against a placebo paste flatters every active; a review that pools against a benchmark fluoride paste, as the 2026 network does, shrinks all the numbers, because fluoride toothpaste is not inert². Reviews also differ in what they count as an outcome, and objective instruments and patients' own ratings do not move together, which is why the Cochrane review found a significant effect on air-blast and tactile scores and none on what patients said⁹. Add a large measured placebo effect⁶ and heterogeneity of 86 to 95% between trials³, and disagreement between honest reviews is the expected result rather than a scandal.