Probiotics for Dental Caries

Do probiotics prevent cavities? Maybe in young children, but the evidence is mixed. A 2023 meta-analysis of 43 randomised trials found fewer non-cavitated and cavitated carious lesions in children (SMD -0.18 and -0.32), and a 2026 meta-analysis rated the evidence moderate-certainty for modest caries prevention in children (RR 0.80, 95% CI 0.68 to 0.95). Yet a 2024 meta-analysis of 19 trials in 2,622 children and adolescents found no effect on the number of carious teeth (mean difference -0.24, 95% CI -0.72 to 0.23; low-certainty evidence), and a 2016 review of 50 studies called the evidence insufficient to recommend probiotics for managing caries.

The reviews that measured it agree on a drop in salivary Streptococcus mutans, a cavity-linked bacterium, which is a marker, not a cavity. The strongest positive trials gave a named strain in milk, or one three-strain tablet, daily or on weekdays to preschoolers. We found no adult benefit, so this page is mainly about children. See dental caries for what is proven (fluoride, sealants, less sugar), and probiotics for the other uses of these supplements.

A cavity forms when cavity-causing bacteria in dental plaque and fermentable carbohydrates, mainly sugars, are out of balance over time (Cochrane). Probiotic trials mostly aim at Streptococcus mutans, one of those bacteria. Probiotics lowered salivary S. mutans counts in a 43-trial meta-analysis (SMD -1.17) and in a 19-trial analysis (-0.40 CFU/mL, moderate certainty), and the drop grew with use: -0.56 at 7 days, -0.81 at 14 days and -0.92 at 28 days.

NIH adds that probiotics may help the body's community of microorganisms return to a healthy state, produce useful substances or influence the immune response. One small SP1 milk trial (42 children) also saw salivary human beta-defensin-3, which its authors treat as a marker of oral tissue homeostasis, fall (597.91 to 126.29 pg/mL); that is a marker, not a mechanism.

Lower S. mutans counts are a stand-in for cavities, and the two do not always move together. A 2014 review found only two articles with caries incidence data, and a 2016 review found S. mutans fell (SMD -1.18) while caries incidence did not significantly (OR 0.60, 95% CI 0.35 to 1.04). In the 9-year follow-up of infants given L. reuteri, there were fewer cavities but no group differences in S. mutans, lactobacilli or salivary IgA.

Lactobacillus counts are murkier still: one meta-analysis found them higher (SMD 1.19), another found no significant change, and a 2026 review warns that lactobacilli are also linked to caries progression in acidic plaque. In a laboratory biofilm model on bovine teeth, L. rhamnosus GG added to S. mutans increased mineral loss. That is not a human result, and GG milk reduced caries risk in a 594-child trial, but it shows the effect is not automatic.

No dose has been established. The trials gave a probiotic as a daily serving or on several days a week, as milk, a chewing tablet or a lozenge, for 3 to 21 months, usually to children aged 1 to 6. Most used one named strain; some used multi-strain products. Daily doses across trials ranged from 10^6 to 10^10 CFU with no clear dose-response relationship, and a 2026 review found no optimal strain, dose or vehicle can yet be recommended.

Several abstracts do not report the dose (the GG milk, SD1 and L. reuteri trials), so check any product's label against a specific trial rather than assuming it matches. Products and strain names vary; a probiotic that names a different strain has no trial behind it for cavities.

In 2023, a meta-analysis of 43 randomised trials in children found probiotics reduced non-cavitated lesions (SMD -0.18, 95% CI -0.3 to -0.06) and cavitated lesions (SMD -0.32, p = 0.0004), and lowered the proportion of children with non-cavitated lesions (RR 0.8, 95% CI 0.67 to 0.97). In 2024, a meta-analysis of 19 trials (2,622 children and adolescents without systemic disease) found no effect on the number of carious teeth (MD -0.24, 95% CI -0.72 to 0.23; I2 = 52%, low certainty) while S. mutans counts probably fell (moderate certainty).

A 2021 analysis of 9 trials in 2,363 preschoolers found median caries incidence of 8.5% on probiotics vs 17.5% on placebo, but judged the certainty low and rated 7 of the 9 trials at high risk of bias. A 2026 meta-analysis of 12 pediatric trials found RR 0.80 (95% CI 0.68 to 0.95) with moderate certainty, and no benefit in 6 adult trials (RR 0.95). They used different outcome measures and trial counts (43 vs 19); the abstracts do not say whether the trial sets overlap or explain why the results differ.

L. rhamnosus SP1 milk is the most-tested positive: in 261 Chilean children aged 2 to 3, new cavitated lesions appeared in 9.7% on SP1 vs 24.3% on standard milk over 10 months, and a further 10-month trial (205 children) found progression to cavitated lesions in 26.8% vs 46.3%. L. paracasei SD1 milk in 487 Thai children cut caries risk by about 17% (IRR 0.83) and roughly doubled lesion regression, and L. rhamnosus GG milk in 594 children lowered the odds of high caries risk (OR 0.56).

Other positives: a three-strain chewing tablet (0.2 vs 0.8 caries increment in 138 children), and L. reuteri in the first year of life (82% vs 58% caries-free at age 9, 113 children). A 2026 systematic review of 12 trials credited SP1 and SD1 in particular and rated certainty moderate for primary prevention and low for secondary outcomes.

A 38-child trial of two L. reuteri strains in children after restorative treatment found 67% recurrence and no group difference, but was underpowered. A 2024 meta-analysis of probiotic beverages found no significant difference in the proportion with cavitated lesions (RR 0.60, 95% CI 0.34 to 1.08) though increments were smaller (SMD -0.23), and it called the evidence not substantial. The lozenge trial with arginine was positive at first (p = 0.007), but its surface-level report found no significant difference (progression 13.6% lower, p = 0.20), and arginine cannot be separated from the probiotics (see arginine for dental caries).

Two positive trials had other changes: the LB21 milk (248 children, 0.9 vs 2.2 dmfs) also contained added fluoride, and an open-label 140-child trial used heat-inactivated L. salivarius, not live bacteria. The L. reuteri follow-up also had a difference in fluoride supplement intake between groups.

Most positive trials come from a few groups: the three SP1 milk trials share authors and two of them share a ClinicalTrials.gov number (NCT01648075), and the two SD1 trials share an author (Teanpaisan R). A 2026 review that highlights SP1 and SD1 is co-authored by an SP1 trial investigator. Dropout was 15% to 25% in several trials. We found no trial in adults showing benefit, no head-to-head strain comparison, and no long-term follow-up beyond the L. reuteri cohort.

In the trials that reported safety, none found serious side effects: no adverse effects in the SD1 trial, no intervention-related adverse events and adherence above 80% in the 205-child SP1 trial, no side effects in the tablet trial, and no serious side effects in the LB21 trial. A 2026 meta-analysis called adverse events mild and infrequent. These trials were small and short, though.

NIH says probiotics have an extensive history of apparently safe use in healthy people, but few studies have looked at safety in detail, harm is more likely in severe illness or weakened immunity, and some products contain organisms not on the label. The abstracts do not say whether children with milk allergy were excluded from the milk trials.

The trial abstracts report no drug interactions. The practical risk is substitution: leaving out fluoride toothpaste or a dentist visit because a probiotic seems to be handling it. Product quality matters too, since NIH notes some probiotic products have contained microorganisms other than those on the label.