Sweeteners are usually judged by one question: do they feed the bacteria that cause cavities today? A quieter question matters over years of daily use: can those bacteria adapt? For two widely used sweeteners the answer — documented in humans — is yes, in two different directions. For almost everything else, nobody has looked.
Research on sweeteners and dental plaque describes three distinct patterns, and knowing which pattern a sweetener belongs to matters more than any single lab result.
Escape resistance (the xylitol pattern). Some sweeteners work like a decoy: cavity-causing bacteria pull the molecule inside, can’t digest it, and waste energy trying. Evolution’s exit is simple — strains that stop importing the decoy survive and spread. The sweetener doesn’t become harmful; its bacteria-suppressing benefit just stops applying to the resistant strains.
Wrong-way adaptation (the sorbitol pattern). Other sweeteners are fermented slowly. With habitual exposure, plaque can get better at fermenting them — producing more acid over time, not less.
Ecological selection (the sugar pattern). Sugar itself doesn’t need bacteria to adapt to it; it is the preferred food. Frequent exposure steadily selects an acid-loving plaque community — the well-described ecological driver of tooth decay.
Xylitol’s dental reputation rests on the decoy mechanism: sensitive Streptococcus mutans imports it through a sugar-transport system and traps it as a dead-end compound, wasting energy. The catch was documented decades ago. In 1987, researchers found that 87% of S. mutans isolates from long-term xylitol users were resistant — they had stopped importing the decoy — versus about 10% in non-users.1 A 1992 follow-up of a Finnish trial found 83% resistant strains in habitual users and 75% in former users years after they stopped, versus 35% in non-users.2 In the lab, fresh clinical strains became resistant within 9–16 serial transfers — a reproducible evolutionary route, not a curiosity.3
Two balancing facts keep this honest. First, resistance does not make xylitol harmful: resistant or not, oral bacteria cannot ferment xylitol to acid, so a resistant mouth reverts toward treating it as an inert sweetener. Second, the clinical evidence for xylitol’s anti-cavity benefit was modest to begin with: the Cochrane review graded the evidence low-quality overall,4 and the largest independent adult trial (33 months, n=691, NIH-funded) found no significant benefit.5 What has never been studied — by anyone — is whether the clinical benefit survives once resistant strains dominate. The resistance literature and the clinical literature have simply never been connected.
More on the sweetener itself: the xylitol reference page.
Sorbitol is fermented by plaque bacteria — slowly, which is why it is far gentler than sugar. But in a four-week human study, daily 10% sorbitol rinses increased sorbitol-fermenting bacteria (including S. mutans) and deepened the plaque pH drop — most sharply in people with low saliva flow.6 Format matters: a chewing-gum study found no community shift, likely because gum stimulates saliva. The pattern is the mirror image of xylitol’s: instead of the sweetener losing its effect on bacteria, the bacteria improve their effect on the sweetener.
More on the sweetener itself: the sorbitol reference page.
For erythritol, sucralose, saccharin, stevia (Reb A), fermentation-made Reb M, and monk fruit, no adaptation study of any design exists — no serial-passage experiment, no habitual-consumer strain survey. Absence of a study is not evidence of safety, and it is not evidence of risk; it is simply an open question. A few of those questions are sharper than others: saccharin is actively pulled into bacterial cells (the same kind of transporter-dependent mechanism bacteria escaped with xylitol), and stevia and monk-fruit sweeteners are glucose-carrying molecules that gut bacteria can demonstrably split — whether any mouth bacterium can learn the same trick has never been examined.
Allulose is the newest molecule on the shelf, and its dental file has three layers. In a small human crossover trial (n=7 adults with high cavity risk, filed in a public FDA docket), a 4.7% allulose rinse left plaque pH statistically indistinguishable from water (minimum 6.43 vs water 6.54) while sugar dropped it to 5.42 — though the study was industry-sponsored and has not been published in a peer-reviewed journal.7 In multi-species laboratory models, allulose supported ~98% less bacterial acid than sugar and preserved beneficial bacterial diversity.8 A third study found a transient pH dip to ~5.4 in an unbuffered cell suspension before recovery — its authors did not consider allulose a cariogenic substrate and noted the dip could reflect trace fermentable sugars in the commercial material, but it is a fair caution against calling allulose completely inert.9 The FDA, after reviewing the evidence, wrote that allulose, “like other non-cariogenic carbohydrate sweeteners…, does not result in a decrease in dental plaque pH below 5.7” and “does not promote dental caries.”10 On adaptation, allulose is exactly like the rest of the unstudied majority: nobody has looked.
More on the sweetener itself: the allulose reference page.
Rankings depend on how you weight the evidence, so this table groups instead of ranks.
| Group | Sweeteners | What’s documented |
|---|---|---|
| Adaptation documented in humans | Xylitol (escape resistance) · Sorbitol (wrong-way adaptation) | Human strain surveys and exposure studies from 1987–1992; persistence for years; lab-reproducible for xylitol |
| No adaptation route identified — but untested | Erythritol · Sucralose | Not meaningfully imported or fermented by mouth bacteria as far as is known; formal adaptation studies: none |
| Open mechanistic questions — untested | Saccharin · Stevia (Reb A) · Reb M · Monk fruit · Allulose | Each has a conceivable route (transporter import, glucose-unit splitting, weak fermentation) that no study has examined |
| The benchmark | Sugar (sucrose) | Adaptation runs the worst direction at every level — the established ecological driver of decay |
The studies that would answer these questions are well understood and mostly cheap: serial-passage experiments (grow plaque bacteria on a sweetener for dozens of generations and watch for adaptation), modern genome sequencing of resistant strains (the existing resistance work predates genomics), and clinical trials that measure whether a sweetener’s benefit survives in mouths where resistant strains dominate. As of this writing, none has been run for any sweetener since the 1990s-era xylitol and sorbitol work.
The resistance findings don’t say that. Chewing gum stimulates saliva, which is good for teeth regardless of sweetener, and xylitol remains non-fermentable — it never becomes food for plaque. What the findings limit is the extra bacteria-suppressing benefit attributed to xylitol itself under long habitual use.
No. Resistant strains simply stop importing it. For those strains xylitol behaves like an inert sweetener — no benefit, no harm. (Xylitol’s serious documented hazard is to dogs, not people.)
“Proven” is stronger than the record. The FDA concluded allulose does not promote dental caries, and the human plaque data behind that conclusion look clean — but they are unpublished docket studies, and the peer-reviewed base is small and new. The honest summary: consistently favorable, not yet deep.
The human studies were published in dental journals in 1987–1992 and were never carried into the big clinical reviews — the Cochrane review of xylitol products never mentions resistance at all. The two literatures have run in parallel for thirty years.