Field guide · Fig. 1–10

Probiotics 101

What’s in the bottle, what happens after you swallow it, and what research can and can’t say yet.

We describe what research found and what labels claim. We don’t tell anyone what to take, or when.

Fig. 1 · The words

Five words, one meaning each

International scientists agreed on what each word means. A product can use the word only if it fits.

Probiotic Live microbes that, in adequate amounts, bring a health benefit. ISAPP · Hill 2014
Prebiotic Something your gut microbes use selectively, with a health benefit. Usually a fibre. ISAPP · Gibson 2017
Synbiotic Live microbes plus what they eat. Either two that each work alone, or two designed to work together. ISAPP · Swanson 2020
Postbiotic Microbes that aren’t alive any more, or their parts, with a health benefit. ISAPP · Salminen 2021
Fermented food Food made by microbes. Not a probiotic unless it carries defined live strains with evidence. ISAPP · Marco 2021

ISAPP is the International Scientific Association for Probiotics and Prebiotics. These are consensus statements: definitions, not findings.

Fig. 2 · Strain, not species

Read the strain, not the species

Effects differ from one strain to the next, even within one species, so the code after the name matters most.

An example label, not a product. CFU means colony-forming units: how many live cells can grow into a colony in a lab.

Genus: Lacticaseibacillus; In 2020 the genus Lactobacillus was split into 25 genera, so old and new labels can name the same microbe differently. Zheng 2020
Species: rhamnosus; Not enough on its own: effects differ strain by strain.
Strain: GG; The code research is about. A species with no strain code can’t be matched to a trial. McFarland 2018
Amount: 10 billion CFU; Experts expect the labelled number to be there at the end of shelf life. Counts fall over time, most when stored at room temperature. ESPGHAN 2017

Across trials Across 228 trials, results differed by strain and by condition. Pooling different strains can mislead. McFarland 2018
Across trials Across 65 trials, where a single strain and a mixture containing it were compared, single strains usually did as well as the mixture. McFarland 2021
Across trials A dose–response showed up for antibiotic-associated diarrhoea. The review found none for most other uses it looked at. Ouwehand 2017 (review)

Fig. 3 · Passing through

Visitors, mostly

In these studies most probiotic strains passed through rather than moved in, and whether one stayed depended on the person.

Seen in people Where strains settled in the gut lining differed by person, by gut region and by strain. Some people’s guts resisted them entirely, and stool tests didn’t show who. Zmora 2018
Seen in people One strain of B. longum stayed at least 6 months in 30% of people, more often in those with little of it to begin with. Maldonado-Gómez 2016
Seen in people After antibiotics, the usual gut community took longer to come back in people given probiotics than in people who recovered on their own. Returning their own stored stool brought it back within days. Suez 2018

Fig. 4 · The journey

From the bottle to the colon

Most of what’s known about capsules, powders and coatings comes from lab models of the gut, not from people.

Mouth Swallowed with or without food
Stomach Acid, the harshest stretch
Small intestine Bile. Where a capsule opens depends on its shell
Colon Where most gut microbes live

In a dish In a lab gut model, capsule-in-capsule designs let more of one bacterium through than a single plain capsule. Reviews say labs overstate protection compared with what’s recovered from people. The capsule maker’s staff were among the authors. Marzorati 2021 Maurya 2026 (review)
Seen in people In 6 volunteers, where a capsule opened depended on its shell and differed a lot between people. They tracked a marker, not bacteria or health. Rump 2021
In a dish Spores of one Bacillus survived stomach acid in a lab gut model and germinated (woke up) further along. No trial we’ve checked shows spore-formers doing better than other probiotics. Keller 2019
Seen in people Live spores of another Bacillus were recovered from volunteers’ stool for up to 12 days after they swallowed them. Ghelardi 2015 (stool recovery)

Fig. 5 · In the bottle

Does the label match the bottle?

Independent labs keep finding gaps between labels and contents. One study found a close match.

Measured in products Only 1 of 16 products with bifidobacteria claims matched those claims in every sample tested. Lewis 2016
Measured in products Over 12 months, counts in commercial multi-strain supplements fell toward the end of their stated shelf life, most when stored at room temperature. Some products missed their label however they were stored. Brousilova 2026
Measured in products In 17 infant products, over 82% of bacterial DNA matched the species on the label. Single-strain products matched best. DeVette 2025
Consensus A European paediatric society’s position paper says departures from label information are common, and expects the labelled amount to be present at the end of shelf life. ESPGHAN 2017
Consensus An expert panel recommends independent third-party testing of quality and label accuracy. Jackson 2019

Fig. 6 · Timing

With food, with fibre, with antibiotics

What research says about when, and with what. It describes studies. It doesn’t tell you what to do.

In a dish In a lab model of the upper gut, more of one product’s bacteria survived when given with a meal or 30 minutes before it, especially with some fat. Meal timing isn’t in any trial in people we’ve checked. Tompkins 2011
Tested in a trial In a 3-week trial, adding a prebiotic (GOS) to a strain gave only a non-significant trend toward higher counts in stool, no more than the strain alone. Krumbeck 2018
Across trials In 19 trials of hospitalised adults, probiotics started within 2 days of the first antibiotic dose went with fewer C. difficile infections than ones started later. Shen 2017

Spacing probiotic and antibiotic doses hours apart isn’t tested in any trial we’ve checked.

Fig. 7 · Transplants

Stool transplants

Moving a donor’s whole gut community has the strongest evidence in this field, for one condition.

Tested in a trial In an open trial that was stopped early, 13 of 16 people with recurrent C. difficile were clear after one donor-stool infusion, against 4 of 13 on the antibiotic vancomycin alone. van Nood 2013
Across trials Across 6 trials of 320 adults, recurrent C. difficile was about 1.9 times as likely to resolve after a transplant. The review found no conclusive evidence on safety. Cochrane 2023
Regulation The FDA has approved two stool-based products to prevent C. difficile coming back in adults after antibiotic treatment: Rebyota (30 November 2022) and Vowst (26 April 2023, capsules). The FDA says Vowst may carry a risk of transmitting infectious agents and may contain food allergens, and that donors and donated stool are tested for transmissible pathogens. FDA · Rebyota approval letter FDA · Vowst announcement
Across trials For ulcerative colitis and IBS, reviews found mixed results and call the evidence weak. Imdad 2023 Lo 2024 Ianiro 2019
Tested in a trial In a 24-adult pilot in people with obesity, weekly donor-stool capsules for 6 weeks showed no significant effect on insulin sensitivity compared with placebo. Yu 2020
Seen in people Two people in separate trials got drug-resistant E. coli bloodstream infections after transplants from the same stool donor. One died. DeFilipp 2019

Fig. 8 · What’s next

The next generation

Gut regulars being tested as products, and why supplements and medicines aren’t the same thing.

Tested in a trial A pilot trial of pasteurised Akkermansia in overweight or obese adults with insulin resistance: insulin sensitivity was higher than on placebo after 3 months (32 adults finished). Safety was the main aim, and it hasn’t been repeated. Depommier 2019
Seen in people A first-in-human study gave a Faecalibacterium strain to 8 adults with Crohn’s disease, for safety. There was no comparison group. Guarino-Vignon 2026
Regulation In the US, the FDA doesn’t approve dietary supplements before they’re sold, and the companies are responsible for their safety. Live bacteria given to treat or prevent a disease go the medicine route instead, which starts with an application to the FDA. FDA · Dietary supplements Q&A FDA · 2023 safety communication

Fig. 9 · Who to ask first

Not for everyone

Research flags some groups.

Across trials A large review of trials found no clear excess of side effects with short-term use. Long-term effects are largely unknown, and the infections reported in case studies were mostly in people already unwell. Gas and bloating are the complaints reviews describe. Hempel 2011
Tested in a trial In severe acute pancreatitis, more people died on a probiotic given by tube than on placebo (24 of 152 against 9 of 144). PROPATRIA · Besselink 2008
Seen in people In intensive care, bloodstream infections with Lactobacillus were genetically matched to the probiotic the patients had been given. Yelin 2019
Across trials Across 60 trials in very preterm babies, necrotising enterocolitis was less common with probiotics. The American Academy of Pediatrics says the evidence doesn’t support routine use for preterm babies. Cochrane 2023 AAP 2021
Regulation In September 2023 the FDA warned health care providers that probiotics can cause serious, potentially fatal infections in preterm babies, after a baby under 1,000 g died of sepsis. Genome sequencing matched the bacterium to the product the baby was given. The FDA says it has approved no probiotic as a drug or biological product for infants. FDA · Safety communication, 29 Sep 2023 FDA · letter to providers

If you or someone you care for is very ill, in intensive care, or a premature baby: worth asking a doctor.

Fig. 10 · Label claims

What labels say, checked

Claims you’ll see on bottles, and what the research shows. No brand is named.

Across trials “Billions means stronger.” A dose–response showed up for one use, antibiotic-associated diarrhoea, and none was found for most others. Ouwehand 2017
Across trials “More strains, more benefit.” Single strains usually did as well as mixtures that contain them. McFarland 2021
No studies found “Rotating strains works better.” Rotating strains isn’t tested in any study we’ve checked. Europe PMC
Seen in people “Reseeds your gut.” Most strains passed through. Staying was the exception, and it depended on the person. Zmora 2018
In a dish “Survives stomach acid, so it works better.” Surviving is shown in labs. No trial we’ve checked shows a better result from it. Maurya 2026
Regulation “FDA-approved.” The FDA doesn’t approve dietary supplements before they’re sold. FDA

Keep exploring