Do Probiotics Work? The Evidence Use by Use, With Numbers

For some uses, yes, and the trials say by how much: taken from the first days of an antibiotic course, live cultures cut the chance of diarrhoea from 19% to 8% in 33 children's trials and by 37% in 11,305 adults.^{1, 2} For IBS the evidence is low certainty, for mood and skin smaller still, and for a stomach bug already under way the largest trials found no difference from placebo. One spoke of the Wellgard probiotics guide: the evidence use by use, as counts.

Evidence at a glance

• Trials: 15 pooled analyses, among them 82 trials in 12,127 people and 7 trials in 1,647.
• Largest single trial: 2,981 UK inpatients aged 65 and over, one multistrain capsule a day for 21 days.³
• Strongest measured outcome: diarrhoea during antibiotics, 8% against 19% on placebo in 33 children's trials.¹
• Not tested: two shop-bought products head to head in any trial we found; a 2021 review counted two trials comparing a single strain with a mixture.²¹

Sources checked on 12 October 2026.

Do probiotics work?

"Work" means an outcome a trial measured; the table gives each use its figure and its certainty.

What conditions do probiotics help with?

In a 2019 Cochrane review of 33 trials in 6,352 children, those given live cultures during an antibiotic course had diarrhoea 8% of the time against 19% on placebo, and 5 billion a day or more did better than lower doses.¹ In 42 adult trials, 11,305 people, the chance fell by 37%.² Across 19 hospital studies, starting within two days of the first antibiotic dose gave a relative risk of 0.32 for C. difficile infection, later 0.70;¹² in the largest UK trial, 2,981 inpatients aged 65 and over given a multistrain capsule for 21 days, 10.8% got diarrhoea against 10.4% on placebo.³ A 2025 Cochrane update of 38 trials rates the C. difficile evidence low certainty.⁴ Timing in full: live cultures with antibiotics.

For colds, a 2022 Cochrane review of 23 trials in 6,950 people taking live cultures for more than three months found a relative risk of 0.76 for at least one upper respiratory infection, low certainty.⁵ In colicky exclusively breastfed babies, pooled trials found 74 minutes less crying a day; a health visitor or GP comes first.¹³

Do probiotics work for IBS and bloating?

A 2023 review of 82 trials in 10,332 people with IBS found a benefit for overall symptoms with moderate certainty for Escherichia strains, low for Lactobacillus strains and very low for combinations; for bloating very low throughout.⁶ NICE advises people who choose to try them to "take the product for at least 4 weeks while monitoring the effect".¹⁴ Strain by strain: which probiotics are best for IBS.

Do probiotics work for constipation?

Across 14 trials in 1,182 adults, live cultures shortened whole-gut transit by 12.4 hours and added 1.3 bowel movements a week, with high attrition and reporting bias.⁷ The most-studied bifidobacterium strain was then not superior to placebo in an eight-week, triple-blind trial.¹⁵

Where did the trials find nothing?

For a stomach bug already under way, the 2020 Cochrane review of 82 trials in 12,127 people found no difference in diarrhoea lasting 48 hours or more in the low-risk-of-bias trials (relative risk 1.00, moderate certainty), and in 971 preschool children five days of one lactobacillus strain did not differ from placebo on severity, 11.8% against 12.6%.^{9, 16} For eczema that already exists, a 2018 Cochrane review of 39 trials in 2,599 people found symptoms 0.44 points lower on a 0 to 20 scale in the 13 trials reporting it, little or no difference.¹⁰ For body weight, 15 trials in 957 overweight adults found 0.6 kg less than placebo, which the authors called small; acidophilus and weight has the weight-gain claim.¹¹ For mood, 34 controlled trials gave small effects on depression (d -0.24) and anxiety (d -0.10) scores, larger in clinical samples; see psychobiotics and mood and, for acne, probiotics and skin.⁸

Do probiotics actually reach the gut?

Yes, alive: in volunteers whose gut lining was sampled while taking an 11-strain capsule, the strains were viable after the stomach, and settling varied by person, region and strain.¹⁷ They pass through rather than stay: in 117 people tracked over eight weeks of one bifidobacterium strain, stool levels rose during dosing and fell back towards baseline after stopping.¹⁸

Does the NHS recommend probiotics?

The NHS page, last reviewed 24 November 2022 and past its review date, says "there's little evidence to support many health claims made about them" and notes some evidence for easing some IBS symptoms.¹⁹ In Great Britain the word "probiotic" is read as a health claim in its own right, and no live culture has an authorised health claim on the GB register, so packs say "live cultures" and state the species and the count as facts.²⁰ The register's refusals turned on strain identity, the definition of a beneficial effect and general-population evidence, not on a finding of no effect, and most trials here postdate them.

How do you run a fair trial of one product?

Pick one outcome you can count (days of diarrhoea, a bloating score), take the product at the label dose daily for four weeks, as NICE advises, or for as long as the trial that measured your outcome: the antibiotic course for diarrhoea, months for colds.¹⁴ More strains is not a quality mark: a 2021 review of 65 trials in 10,863 people found single strains and mixtures containing the same strain equivalent in most cases.²¹ If nothing has changed by the end, stop; how long they take to work gives the measurement points.

Who should take care?

In pooled trials, side effects were no more common than on placebo: relative risk 0.86 across 37 antibiotic trials, wind and looser stools the usual early reports.⁴ The exceptions are people with a weakened immune system, people who are seriously ill or in hospital with a central line, premature babies, and anyone whose doctor has said so; case reports describe serious infections in severely ill children with central lines.¹ In a 2021 Cochrane review of trials that gave live cultures in pregnancy to lower the chance of gestational diabetes, pre-eclampsia was more common in the women who took them (relative risk 1.85, high certainty), so speak to a midwife or GP before any supplement in pregnancy.²²

What we don't know

• No pooled analysis reports what happens to an outcome after stopping, beyond stool levels of the strain.¹⁸
• No trial has randomised when a culture starts against an antibiotic course; the two-day finding is a meta-regression.¹²
• No trial has tested whether a count labelled at manufacture and one guaranteed to the best-before date give different results.

Frequently asked questions

What problems do probiotics help?

Counted in trials: diarrhoea during an antibiotic course, C. difficile diarrhoea, colds over months of daily use, infant colic in breastfed babies, and, at low certainty, IBS symptoms and constipation.^{1, 2, 5, 6, 7, 12, 13}

Can you fix a leaky gut with probiotics?

"Leaky gut" is not a diagnosis in UK practice; intestinal permeability is a laboratory measure, and no trial has shown a live-culture capsule changing a symptom that way.

Are probiotics worth the money?

It depends on the use: across 42 trials in 11,305 adults, people who took live cultures during an antibiotic course were 37% less likely to get diarrhoea; for a stomach bug already under way, the largest trials found no difference from placebo.^{2, 9}

Do gut supplements actually work?

Fibre has the better-replicated record: in 16 trials in 1,251 constipated adults, 66% responded against 41% on control, psyllium only above 10 g a day; what psyllium husk does has the doses.²³ For live cultures the table is the answer, and how to choose a probiotic turns it into a label check.

Sources

1. Guo Q, et al. Cochrane review, 33 randomised trials, 6,352 children. 2019. PubMed 31039287.
2. Goodman C, et al. Pooled analysis, 42 randomised trials, 11,305 adults. 2021. PubMed 34385227.
3. Allen SJ, et al. PLACIDE. Randomised trial, 2,981 UK inpatients aged 65 and over, 21 days. 2013. PubMed 23932219.
4. Esmaeilinezhad Z, et al. Cochrane review, 38 randomised trials, 13,179 people. 2025. PubMed 40931979.
5. Zhao Y, et al. Cochrane review, 23 randomised trials, 6,950 people. 2022. PubMed 36001877.
6. Goodoory VC, et al. Pooled analysis, 82 randomised trials, 10,332 people with IBS. 2023. PubMed 37541528.
7. Dimidi E, et al. Pooled analysis, 14 randomised trials, 1,182 adults with constipation. 2014. PubMed 25099542.
8. Liu RT, et al. Pooled analysis, 34 controlled trials. 2019. PubMed 31004628.
9. Collinson S, et al. Cochrane review, 82 trials, 12,127 people. 2020. PubMed 33295643.
10. Makrgeorgou A, et al. Cochrane review, 39 randomised trials, 2,599 people with eczema. 2018. PubMed 30480774.
11. Borgeraas H, et al. Pooled analysis, 15 randomised trials, 957 overweight adults, 3 to 12 weeks. 2018. PubMed 29047207.
12. Shen NT, et al. Meta-regression, 19 studies, 6,261 hospital adults. 2017. PubMed 28192108.
13. Vaz SR, et al. Pooled analysis of randomised trials, infants under three months. 2024. PubMed 37962097.
14. NICE. Irritable bowel syndrome in adults: diagnosis and management (CG61). Published 23 February 2008, last updated 4 April 2017. nice.org.uk.
15. Ala-Jaakkola R, et al. Randomised triple-blind trial, adults with constipation, four centres, 8 weeks. 2025. PubMed 40320938.
16. Schnadower D, et al. Randomised trial, 971 children with acute gastroenteritis, 5 days. 2018. PubMed 30462938.
17. Zmora N, et al. Clinical study with endoscopic sampling, healthy volunteers. 2018. PubMed 30193112.
18. Charbonneau D, et al. Randomised trial, 117 people, 8 weeks. 2013. PubMed 23549409.
19. NHS. Probiotics. Page last reviewed 24 November 2022 (review overdue). nhs.uk.
20. DHSC. Great Britain nutrition and health claims register. Published 31 December 2020, last updated 19 May 2026. gov.uk.
21. McFarland LV. Systematic review, 65 randomised trials, 10,863 people. 2021. PubMed 32274669.
22. Davidson SJ, et al. Cochrane review, 7 randomised trials, 1,647 pregnant women. 2021. PubMed 33870484.
23. van der Schoot A, et al. Pooled analysis, 16 randomised trials of fibre, 1,251 adults. 2022. PubMed 35816465.

Spotted something out of date? Email support@wellgard.co.uk with the page address and the sentence.