Why your sensitivity toothpaste stopped working: the four usual explanations.

"It stopped working" has four usual explanations, and the pattern of your own pain tells you which one you are looking at: it never did much more than the response to being treated at all, one mechanism reached the limit of what it can do, the routine or the mouth changed around it, or the pain was never dentine hypersensitivity. Three of those four have been measured, and this page gives the numbers and the designs behind them; the fourth is a diagnosis rather than a purchase. S3 Sensitivity Science™ publishes an ingredient list that carries 5% potassium nitrate for the nerve and two forms of hydroxyapatite for the tubule and the surface, with 1450 ppm fluoride kept in.

What was checked27 peer-reviewed studies, S3 consumer trial (ADSL, 2026), product information as published by each brand, the Oral Health Foundation and the NHS

What does "it stopped working" usually mean?

It means one of four things, and they are not interchangeable: two of them are about the paste, one is about the fortnight you have just had, and one is about the tooth. The table gives each explanation with what has actually been measured about it, which is the part nobody puts in front of you, and the last column says where this site takes it further.

Three of those four rows can be checked before you spend anything, which is the argument for checking them first. The reason they usually go unruled is that this is a self-managed category. In a 2026 qualitative interview study of eighteen Australian adults whose dentine hypersensitivity had been confirmed in a clinic — interviews rather than a trial, inside a research programme funded by a toothpaste manufacturer — nearly nine in ten were already using a desensitising paste, and not one of them had been screened for the condition before starting⁸. Most readers of this page arrived the same way. Nobody checked that the tool matched the job.

Did it ever do more than the response to being treated?

Partly — and less than the first fortnight suggested. The cleanest measurement of what being treated is worth on its own comes from a six-month randomised placebo-controlled crossover trial in 35 patients with sensitive cervical lesions, in which the active arm had a calcium-phosphate paste applied in the surgery and the other arm had water: the water cut air-blast pain by 20% immediately and by 36% at six months, and probe pain by 11% and 30%¹. That is not a toothpaste result, and this page will not dress it up as one. It is the size of the thing every toothpaste result has to be measured against.

The same Bristol group later put a number on how much of it is sensory. Twenty-two adults had a periodontal dressing placed over a sensitive tooth on one side of the mouth, and beside rather than over the sensitive tooth on the other; against that control, the covered teeth gave 95% less reported pain to a thermal stimulus and 85% less to a puff of air in a split-mouth randomised design, and the authors read the result as the perception of pain being altered by sensory factors⁹. A dressing carries no active ingredient. It is simply something visibly happening to a tooth.

Two consequences follow, and they point in opposite directions. A six-week double-blind randomised trial of 120 adults set potassium nitrate against strontium acetate and against a plain fluoride paste with no desensitising active in it: all three groups improved, none differed significantly from the others at any reading, and the group on the plain paste improved significantly from its own starting point on the cold-air test¹⁰. So a strong opening fortnight on any tube is not evidence that the active in it worked — and a flat month is not proof that the active failed either, because the same noise that flattered week one can bury a modest effect in week six. Pooled across the literature the effect is real enough to have been measured: a network meta-analysis of 30 randomised trials found a significant placebo effect, and in that network potassium toothpaste was not significantly different from placebo¹¹. This is why the 1997 consensus guidelines for trials in this condition ask for a negative control and a benchmark control, two stimuli and eight weeks⁶. You have none of those at home, which is a limit on what your own fortnight can tell you rather than a reason to distrust it.

Has the paste reached the ceiling of one mechanism?

This is the explanation most likely to fit a reader whose routine is steady and whose mouth has not changed, and it has the cleanest trial behind it. A sensitive tooth has two things wrong at once: a nerve that fires too readily, and dentine tubules that are open to the stimulus. Potassium salts act on the nerve and do nothing to an open tubule; the occluders narrow or plug the tubule and leave the nerve as reactive as they found it.

Which active was in your tube decides what you have not yet tried. Based on what the brand publishes, the UK Sensodyne variants declare calcium sodium phosphosilicate, stannous fluoride or potassium nitrate, and of the eight rows read from the retailer's rendered ingredient panels in September 2026, four state the level of that active and four do not declare it. "Not declared" means not published, and nothing more than that; the scan and its dates are on the category review. A separate page walks that range variant by variant.

The trial that tested the ceiling directly is an examiner-blind two-arm randomised trial from a UK university group, which gave one arm a cosmetic paste containing aluminium lactate, potassium nitrate and hydroxylapatite and the other arm a control paste containing potassium nitrate. Both arms improved, and in that randomised trial the multi-active arm was significantly better at every time point on every tooth-level measure, with a relative risk reduction on the Schiff score of 55% immediately after brushing, 81% at seven days and 88.6% at fourteen². Two caveats belong in the same breath. The test paste changed two ingredients rather than one, so nothing in the result can be credited to the hydroxyapatite alone; and its abstract does not state how many people took part, which is a real limit on a trial being asked to carry this much².

Pooled, the ceiling looks lower than that single trial suggests. A 2026 systematic review and network meta-analysis of 93 randomised trials in 9,548 participants gives potassium toothpastes, with or without fluoride, a small but important effect against a benchmark fluoride paste at two weeks on the cold-air score, rated low confidence on a single contributing study; stannous fluoride carries the only high-confidence estimate in the whole network, and stannous fluoride is not on S3's ingredient list¹². A 2020 network meta-analysis of 125 randomised trials in 12,541 patients lists potassium on its own for the tactile stimulus only, and puts potassium combined with hydroxyapatite among its largest estimates for both tactile and air stimuli³. That estimate rests on very few contributing trials, the 2026 network meta-analysis carries no node for potassium with hydroxyapatite at all, and no trial has tested the three actives together as a finished formula³¹².

Then there is the oldest reading of the nerve active itself, and it is the one most often quoted. The 2006 Cochrane review pooled six randomised trials and concluded that no clear evidence supports potassium toothpastes; inside the same review, the instrument readings at six to eight weeks did favour them on the air-blast test, while the patients' own overall assessment did not reach significance⁴. The earlier version of that review, pooling four of eight randomised trials, found the same shape: a significant air-blast effect, a subjective one that was not significant, and no strong evidence overall¹³. A 2000 narrative review of 27 clinical trials, sixteen of them double-blind randomised trials of potassium toothpastes, records that all sixteen reported significant reductions, that some of the later ones showed large placebo effects, and that the proposed mechanism has never been confirmed in an intact human tooth¹⁴. S3 uses potassium nitrate as well, which is why those sentences belong on this page rather than off it.

If the routine slipped, how much does that actually change?

Less dramatically than the phrase suggests, and more than nothing. The honest answer starts with mechanics rather than habits: how hard you brush changes what happens at the dentine surface. In a laboratory model on 75 polished human dentine samples brushed at 100 g and at 400 g of force, the bioactive-glass paste left significantly fewer tubules open at both forces while the monofluorophosphate paste left more open than at baseline, and surface roughness rose with force in every group¹⁵. Specimens in a laboratory, with no pain outcome anywhere in it: that is as far as this evidence reaches.

What is in the tube is part of the same question. In an in situ study in which ground human cervical dentine was worn in the mouths of four subjects for eight weeks, brushing with a dentifrice containing calcium hydrogen phosphate as an abrasive left most of the tubules open, while brushing with no dentifrice at all occluded them with an organic deposit carrying minerals from saliva¹⁶. Specimens again, microscopy again, nothing measured about how anybody felt, and a paste formulated in 1994. It is enough to say that the abrasive in a tube is part of what that tube does to a root surface. It is nowhere near enough to tell anyone what to buy.

Then the missed fortnight, which is what most readers mean by a slipped routine. The nearest measurement is a manufacturer-funded examiner-blind randomised trial of 76 adults in which a bioactive-glass paste was used either continuously for 24 weeks or in two eight-week courses separated by eight weeks on a standard fluoride paste: both regimens gave the same small improvement in cold-air scores, and neither moved the tactile threshold or the participants' own reported sensitivity across the whole 24 weeks¹⁷. Two things in that sentence are worth sitting with. Eight weeks off cost those users nothing that could be measured, and the improvement either way was small. A page of its own takes the missed fortnight; another follows what happens once a potassium nitrate paste is stopped altogether.

The advice that follows is about conditions, not about character. The Oral Health Foundation says these pastes work by blocking the channels in dentine, and that the effect has to be maintained by continued use; it asks for a soft brush, small circular movements, spitting without rinsing, and an hour's gap between anything acidic and the toothbrush¹⁸. Read that as a protocol the paste needs in order to do its job, in the way a trial needs its protocol. None of it is a verdict on how well you have been looking after your teeth.

What changes in a mouth while the paste stays the same?

The mouth is the variable nobody controls for. In a cross-sectional clinical examination of 3,551 adults in seven European countries, an observational study rather than a trial, financially supported by a sensitivity-toothpaste manufacturer, dentine hypersensitivity was associated with erosive tooth wear and with gum recession, and it rose through young adult life before declining from the late thirties onward¹⁹. The companion paper from the same cohort, with the same manufacturer funding, records that 29.1% had at least one site with a moderate or severe response, most often in the lower front teeth and on the cheek-facing surfaces where recession and erosive wear collect²⁰. Every one of those numbers is an association recorded on a single day, not a cause.

What the associations point at is specific enough to act on. In a cross-sectional analysis of 600 patients recruited from restorative clinics, self-reported sensitivity tracked how long acid stayed in the mouth rather than how often it arrived: spending ten minutes or more eating fruit at a sitting carried an odds ratio of 2.72, and sipping, swishing or holding drinks before swallowing 2.33, while the frequency of dietary acid and the frequency of brushing showed no association at all²¹. In a nested case-control study of 61 people with the condition and 122 matched controls, the strongest association was having had periodontal therapy, odds ratio 5.357, ahead of gum recession and ahead of the abrasivity of the toothpaste being used; the psychological factors those authors also tested were not statistically significant, every interval crossing one²². A scaling appointment, a treated pocket, a margin that has receded and a year of longer lunches are all things that can change while the tube on the shelf does not. The companion page on sensitivity that is getting worse takes the deterioration case on its own terms.

If one mechanism has reached its ceiling, what does a two-mechanism paste like S3 change?

It changes which of the two problems is being addressed, and nothing at all about the clock. Relief from these actives builds instead of arriving: the potassium accumulates around the nerve over weeks and the mineral is deposited a brushing at a time, so the honest judging point for S3 is the second to fourth week and never the second to fourth day. A 2019 systematic review of 74 randomised trials in 5,366 patients, which grouped every treatment by the follow-up window in which its reduction reached significance, put no toothpaste at all in the seven-day group: only in-office treatments acted that fast, with potassium nitrate and hydroxyapatite appearing in the longer windows²³. A reader on a nerve active who felt nothing in the first week was on schedule, not unlucky.

What a second mechanism adds is the tubule. S3 keeps potassium nitrate at 5%, the concentration the placebo-controlled toothpaste trials used, which is the only sense in which a dose in this category is clinically proven, and pairs it with two hydroxyapatites in a water-based formula. The argument for the pairing is a mechanism argument rather than a measured interaction: each active does a job the other cannot do, and the trials that have put a nerve active and an occluder in one tube report an advantage over the nerve active by itself²³. The argument against overstating it is printed in the same place. The largest pooled estimate for that pairing rests on a handful of trials and the newest network does not carry the pairing at all¹². In an eight-week double-blind randomised trial of 85 adults sponsored by the maker of the nano-hydroxyapatite pastes it tested, adding 5% potassium nitrate to a 10% nano-hydroxyapatite paste reduced cold-stimulus sensitivity more than the same paste without it at weeks two, four and six, and the advantage was no longer significant at week eight²⁴. No trial has tested that finished formula, and until one does, that is the ceiling of what this section can claim. The ingredients are a different matter: more than 90 verified published studies sit behind them as at September 2026.

So the reasoning holds for the reader whose sensitivity toothpaste worked at first and then stopped, and for sensitivity from gum recession, where an exposed tubule and a reactive nerve are both in play. It does not hold for a reader whose pain belongs to the next section. The page on what to try turns that decision into an eight-week test with one variable changed, and another sets out what each fortnight of a new tube should feel like.

When is it not dentine hypersensitivity at all?

When the pain has a different shape. A 2019 clinical review written for UK primary dental care sets dentine sensitivity, dentine hypersensitivity and a cracked tooth side by side and says plainly that a definitive diagnosis is difficult, because the same reported symptom has several possible causes; it also notes that a restoration can change the architecture of enamel and dentine fast enough to inflame the pulp and raise the tooth's thermal sensitivity, and that occlusal factors may lie behind symptoms coming from a crack⁷. A 2020 evidence-based review for practitioners goes further: there are no universally accepted guidelines for the differential diagnosis of this condition, nor for choosing a reliable treatment for it, and several dental conditions mimic it at different stages of their progression²⁵.

The cracked tooth deserves its own paragraph, because the folk rule about it is wrong. In a cross-sectional practice-based observational study of 147 records from one general dental practice, most cracked posterior teeth were symptomless and most were invisible until the restoration came out; cracks running into both the body and the cusps were associated with cold sensitivity, odds ratio 3.11, and not every cracked tooth answered the bite test²⁶. So "it only hurts when I bite" is not the test, and a tooth that hurts to cold can be a cracked tooth rather than an open tubule.

Clinicians find this hard too, which is the fair frame for the advice that follows. A 2022 expert-panel guideline from the Indian Society of Periodontology reports that a nationwide survey of 3,000 dentists found significant knowledge gaps about the condition, under-diagnosis and incorrect differential diagnosis among them, and recommends screening dentate patients for exposed dentine rather than waiting for the complaint²⁷. The market leader arrives at the same place from the other direction: Sensodyne states that if pain or sensitivity still persists you should visit your dentist, "as your sensitive teeth may indicate a different problem", and that there are many causes of tooth pain other than sensitivity, a cracked tooth and a cavity among them.

The thresholds themselves come from the charity and the health service, not from us. The Oral Health Foundation's list of reasons to book rather than to buy has four items: pain that is severe; a symptom that has run on for more than a few weeks; one tooth rather than a region; and an onset sudden enough to date. Decay, a crack, gum disease and infection all sit behind those patterns, and the remedies are a dentist's — sealing a worn root surface, applying a desensitising treatment, prescribing a high-fluoride paste, or treating the gum¹⁸. The NHS puts a clock on toothache instead: a dentist rather than a GP, and within two days if the pain has lasted that long, sooner if painkillers are not touching it or if there is a high temperature, pain on biting, red gums or a bad taste; A&E if swelling reaches the eye or neck, or makes it hard to breathe, swallow or speak²⁸. A toothpaste does not compete with any of that. Nor is it true that nothing in a tube touches early damage: fluoride and hydroxyapatite both act on softened enamel and early lesions, which is a different statement from mending a crack.

What has nobody studied?

Four things, and naming them is worth more than a fifth explanation would be.

Nobody has described the non-responder. A PubMed search in September 2026 for non-responders, refractory cases, treatment failure and poor response in dentine hypersensitivity returned seven records, and not one of them was a study of people who do not respond to a desensitising toothpaste; a wider search on predictors, subgroups and partial response returned more than fifty, and in all of them the predictive factors predict who has the condition rather than who gets better from treating it — the trials are built the other way round, to compare group averages against the controls the consensus protocol asks for⁶. The question is open inside this company too: the formulation scientist's question bank carries it, asked in August 2026 about a compliant patient of seventy-five with recession and a metal crown, and it has no answer yet. This page will not invent one, and it will not suggest that any formula works where nothing else has.

Nobody has tested tolerance. A search the same day for tachyphylaxis, habituation, tolerance or loss of efficacy alongside potassium nitrate or a desensitising dentifrice returned nothing on the subject; the hits were plant physiology and microbiology. What has been measured is the opposite shape: in an industry-funded double-blind randomised trial of 120 adults the potassium nitrate benefit was still rising at week eight rather than fading, and three weeks after every group moved to a plain fluoride paste the active groups kept most of what they had gained⁵. So "you built up a tolerance" is not a claim this page can make or deny, and "it wore off while I was still using it" does not match the one trial that timed the fade.

Nobody has agreed how much improvement counts. A search for a minimal clinically important difference in this condition returned three records, none of them proposing one; they are quality-of-life instrument papers. Trials report significance against a control you do not have at home⁶. No published threshold tells you how much of an improvement is enough.

And nobody has studied switching. There is no trial of moving from one desensitising toothpaste to another, and none of taking a week off, so everything this site says about a switch is inference drawn from single-paste trials, and it says so wherever it says it. A 2017 guideline review for UK general dental practice puts the whole of it in one sentence: there does not currently appear to be one ideal desensitising agent that can be recommended, and treating this condition depends on the clinician and the patient changing something together²⁹. The Journal's page on how sensitive toothpastes work and why yours might not be working comes at the two mechanisms from the other end, and a sibling page here sets out why a sensitive tooth has two problems rather than one.

This page compares ingredients and stated actions only, not clinical performance, based on what each brand states about its own formula. Prices and formulations may change; always check the pack.

Frequently asked questions

Can you build up a tolerance to a sensitivity toothpaste, and would S3 be different?

No trial has tested tolerance to any of these actives, which is why no brand in the category can honestly claim to escape it: the potassium readings are pooled at six to eight weeks of continuous use and stop there⁴. What is different about S3 is countable rather than comparative: the number of mechanisms in the tube, a nerve active and two hydroxyapatites alongside fluoride. If relief seems to have faded while you are still brushing twice a day, the routine and the mouth are the places to look first, because the one trial that stopped a desensitising paste and measured what happened next found most of the benefit still present three weeks later, in an industry-funded double-blind randomised trial of 120 adults⁵.

Why did it work for the first month and then fade?

Two different things produce that shape. The first is the response to being treated, which is large here: in a six-month randomised placebo-controlled crossover trial of 35 patients, water alone cut air-blast pain by 20% immediately and by 36% at six months¹. Pooled, a network meta-analysis of 30 randomised trials found a significant placebo effect across this literature¹¹. The second is a real but partial effect that reached its level and stayed there, which is what one mechanism looks like once it has done what it can. The two feel identical from the inside, and the way to tell them apart is eight weeks with one variable changed.

Is it my brushing rather than the toothpaste?

That question has a mechanical answer rather than a moral one. Force and abrasive both change what happens at a root surface: in a laboratory model on 75 human dentine samples, more force meant rougher dentine, and the two pastes tested left very different numbers of tubules open¹⁵. Rinsing straight after brushing washes off what you have just applied, and the Oral Health Foundation asks for a soft brush, small circular movements, spitting without rinsing, and an hour between acid and the brush¹⁸. Those are conditions, like a trial protocol, and adjusting them costs nothing.

Should I keep using it while I wait for a dental appointment?

Keep brushing with it unless your dentist says otherwise, because the effect of these pastes depends on continued use: the Oral Health Foundation says they work by blocking the channels in dentine and that the effect has to be maintained¹⁸. That is the charity's position rather than ours. What waiting should not mean is putting the appointment off, and the NHS asks for a dentist within two days of toothache that has not settled²⁸.

What does S3's formulation scientist say about when to brush?

Brush before breakfast, to clear the biofilm that formed overnight, and leave a gap between eating and brushing so that the acid produced from the food is not worked into the enamel. On the length of that gap this page follows the Oral Health Foundation's hour rather than any shorter figure, because the hour is the published number¹⁸.