A probiotic is a live microorganism that is supposed to do you good when swallowed in sufficient numbers. The category is enormous and the products in it have almost nothing in common with each other: a yeast, a single bacterial strain, or a blend of a dozen, in capsules, drinks and yoghurts, at doses spanning orders of magnitude.

That is the single most important thing to understand before reading any probiotic trial. A result for one strain at one dose in one condition is a result for that strain, that dose and that condition. The reviews that pool across strains get averages that may describe no product on any shelf, and the best reviews say so.

Proposed mechanisms include competing with harmful bacteria for space and nutrients, producing acids and antimicrobial compounds, strengthening the gut lining, and signaling to the immune system through the gut wall. For antibiotic-associated diarrhea the logic is direct: antibiotics clear the normal flora and let opportunists, including Clostridioides difficile, take over, and a probiotic taken alongside is meant to hold the ground until the flora recovers. Most probiotic strains do not colonize permanently; they pass through, which is why effects stop when the product does.

Antibiotic-associated diarrhea: established. A 2012 meta-analysis in JAMA of 63 randomized trials in 11,811 people found probiotics reduced the risk of diarrhea on antibiotics by about 42%, a number needed to treat of 13. A 2021 meta-analysis of 42 trials in 11,305 adults found a 37% reduction with moderate-certainty evidence, larger at higher doses and with certain Lactobacillus and Bifidobacterium species, and no benefit in trials where the baseline risk was already low. The 2017 Cochrane review on C. difficile diarrhea specifically found probiotics reduced it, with the clearest benefit in people at higher baseline risk.

Irritable bowel syndrome: strain by strain, low certainty. A 2023 meta-analysis in Gastroenterology of 82 trials in 10,332 patients found moderate-certainty evidence for a benefit of Escherichia strains on global symptoms, low certainty for some Lactobacillus strains, and very low certainty for most combinations. Only 24 of 82 trials were at low risk of bias. Probiotics were not associated with more adverse events than placebo.

General health in healthy people. Immunity, mood, weight, skin: trials exist for individual products and are small; no pooled evidence supports taking a probiotic for general wellness.

In healthy people and in the antibiotic trials, probiotics caused no more adverse events than placebo. The serious risks are in specific groups: bloodstream infections with probiotic organisms have been reported in critically ill, immunocompromised and premature patients, and in people with central venous catheters. Saccharomyces fungemia in intensive care is the best-documented. For those groups a probiotic is a clinical decision. Bloating and gas in the first days are common and usually settle.

Probiotics are sold as foods and dietary supplements in every major market, with no review of claims before sale, and the European food claims authority has rejected essentially every probiotic health claim submitted to it, which is why European products describe their contents rather than their effects. In the United States the FDA regulates them as foods or supplements unless marketed to treat disease. No probiotic is a licensed medicine in the US, UK or EU. Viable counts at the point of sale, rather than at manufacture, are the quality problem independent testing most often finds.