How to choose a probiotic: what the evidence actually requires
Walk into any pharmacy and the probiotic shelf presents an overwhelming array of options. Products compete on CFU counts in the tens of billions, on the number of strains, and on claims spanning digestive health, immune support, mood, skin, and more. The diversity is impressive. The clinical basis for most of it is harder to establish.
Choosing a probiotic effectively requires a different framework from what most marketing implies. The evidence that supports probiotic use in specific contexts is real, but it is narrow, strain-specific, and condition-specific in ways that most consumer products actively obscure. Understanding what the evidence actually requires is the starting point for making a sensible choice.
Start with the condition, not the product
The most common mistake in probiotic selection is choosing a product and then hoping it addresses a health concern, rather than identifying what the evidence supports for a specific concern and then finding a product that matches it.
Clinical trials of probiotics are not conducted on categories of products. They are conducted on specific strains, at specific doses, for specific conditions, in specific populations. A meta-analysis finding that Lactobacillus rhamnosus GG reduces the duration of antibiotic-associated diarrhoea does not tell you anything about what a multi-strain supplement containing an unspecified L. rhamnosus will do in that context. The evidence applies to the strain studied, in the condition studied.
The practical implication is that before evaluating any product, you need to be clear about what you are trying to address. The answer to that question determines which strains have relevant evidence and, by extension, which products deserve consideration. A person taking a course of antibiotics who wants to reduce their risk of associated diarrhoea is looking for a different evidence base than someone managing IBS, an infant with colic, or a woman trying to support vaginal microbiome health. These are not interchangeable contexts, and the evidence for each points to different strains.
Why the strain designation is non-negotiable
Probiotics are identified at three taxonomic levels: genus, species, and strain. A full strain designation looks like this: Lactobacillus rhamnosus GG. Lactobacillus is the genus, rhamnosus is the species, and GG is the strain identifier. Clinical trials are conducted at the strain level. Benefits established for one strain do not transfer to other strains, even within the same species.
A systematic review and meta-analysis published in Frontiers in Medicine examined 12 meta-analyses across multiple probiotic strains and conditions and found strong evidence that probiotic efficacy is both strain-specific and disease-specific. This is not a minor academic footnote. It means that a product labelled "Lactobacillus rhamnosus" without a strain identifier cannot be evaluated against the LGG evidence base, regardless of how similar the name appears.
Most consumer probiotic products do not list the full strain identifier. Some list genus and species. Many list genus alone, or use proprietary blend language that provides no useful information. A product cannot be assessed against the clinical evidence if the strain is not identified.
This is the first and most important filter. If a product does not identify its strains to the level of the strain designator, it cannot be compared with published trial evidence. That does not automatically mean the product is ineffective, but it does mean there is no basis on which to predict its effects from the literature.
CFU count: what it means and what it does not
CFU stands for colony-forming units, a measure of viable microbial cells. Consumer marketing has drifted steadily towards higher CFU claims. Products now commonly advertise 50 billion, 100 billion, or more CFU per dose. The implication is that higher CFU means more potent or more effective. The evidence does not consistently support this.
Effective doses in clinical trials vary considerably by strain and condition. Well-powered trials demonstrating meaningful clinical outcomes have used doses ranging from around 1 billion to 100 billion CFU. Higher doses are not consistently associated with greater effectiveness, and for some applications, moderate doses used in the relevant trials are the appropriate reference point rather than the highest commercially available count.
A more clinically relevant question than the CFU count is whether the stated CFU is guaranteed at expiration. Bacteria die over the shelf life of a product, and a supplement stating "50 billion CFU at time of manufacture" offers no assurance about how many viable organisms remain at the point of use. The meaningful figure is CFU at expiration. Products that do not specify this are making a claim that is effectively unverifiable by the consumer.
Refrigeration requirements are also relevant. Some strains require cold storage to maintain viability; others are sufficiently shelf-stable in appropriate encapsulation. A product requiring refrigeration that has been stored or shipped at ambient temperature may not contain what the label claims.
The delivery problem: surviving the stomach
The gastrointestinal tract is a genuinely hostile environment for many microorganisms. Gastric acid, bile, and pancreatic enzymes present successive challenges that many bacterial strains do not survive in sufficient numbers to colonise the small intestine, which is where most of the relevant action is for gut health applications.
This is not a trivial consideration. In vitro studies showing that a strain tolerates acid conditions are not the same as clinical evidence that it survives in adequate numbers in the human gut after oral ingestion. The two do not always correspond, and clinical efficacy is ultimately the relevant measure.
Delivery mechanisms designed to protect bacteria through the stomach include acid-resistant enteric coatings, nested capsule systems, and moisture-resistant encapsulation. Whether these improve outcomes depends on the strain: some have inherent acid tolerance, while others are more susceptible to gastric conditions. It is also worth noting that where a strain has demonstrated clinical efficacy in RCTs using standard oral capsules, delivery was by definition adequate in those trials. The clinical evidence already incorporates whatever survival rate was achieved. Delivery mechanism considerations are most relevant when evaluating products that differ substantially in formulation from those used in the underlying trials.
Saccharomyces boulardii, being a yeast rather than a bacterium, has a natural tolerance for gastric acid that bacterial strains generally lack. This is one of the reasons it performs well in conditions where oral bacterial probiotics struggle to deliver sufficient viable cells to the relevant site.
Multi-strain products: more is not necessarily better
Multi-strain probiotic supplements are marketed on the logic that a wider range of strains replicates the diversity of a healthy microbiome more closely than a single strain, and that diverse inputs produce broader effects. This reasoning sounds intuitive. The evidence base for it is weak.
For any condition where specific strain-level evidence exists, a well-characterised single-strain product matching that evidence is generally a more defensible choice than a broad-spectrum supplement. Adding strains with no evidence for the intended condition does not improve the product's effectiveness for that condition. It may dilute the dose of the relevant strain, and strains have been hypothesised to compete for colonisation in ways that reduce the effective presence of any individual organism, though direct human evidence for this mechanism is limited.
Multi-strain products may be appropriate when the clinical goal is genuinely broad, for example general microbiota restoration following extended antibiotic use, where no single strain has comprehensive evidence. Even here, the relevant question is whether the included strains are each characterised and each have relevant evidence, not simply whether the total count is high.
The practical advice is to treat strain count as a neutral feature rather than a positive one, and to prioritise the presence of strains with evidence for the specific application over the total number of strains listed.
Reading the label: a practical checklist
A probiotic product worth evaluating should allow you to answer the following questions from the label or the manufacturer's product information.
What is the full strain designation, including the strain identifier? If the label lists only genus or genus and species, the product cannot be assessed against strain-specific trial evidence.
What is the CFU count at expiration, not at manufacture? If only the manufacture count is stated, this is a meaningful limitation.
Is the CFU count per dose or per capsule, and what is the recommended daily dose? These are sometimes different, and dosing calculations should use the correct figure.
Is the product third-party tested by a named certifier such as NSF International or USP? "Third-party tested" without naming the certifier is not a verifiable claim.
What are the storage requirements, and have they been maintained throughout the supply chain? Refrigeration-dependent products that have been shipped at ambient temperature may not be viable.
What the evidence does and does not support
There is meaningful clinical evidence supporting specific probiotic strains for specific indications. Antibiotic-associated diarrhoea prevention has the most replicated evidence base across multiple strains, particularly LGG and S. boulardii, though effect sizes vary across trials and guideline bodies have reached different conclusions on how to weigh the heterogeneity, some recommending cautious use, others against routine use outside specific contexts. Acute infectious diarrhoea treatment in children has meaningful evidence for several strains including LGG, S. boulardii, and Lactobacillus reuteri DSM 17938, and has attracted more supportive guideline positions in the paediatric context. Certain IBS applications have signal for specific strains, though certainty is low, placebo response rates are high, and replication across trials is inconsistent.
Evidence for broader wellness claims, general immune support in healthy adults, mood support, energy, skin health, weight management, is considerably weaker. A 2026 systematic review and meta-analysis found no significant effect of probiotic supplementation on gut microbiota alpha-diversity in healthy populations. This addresses specifically the claim that probiotics increase microbiome diversity. It does not rule out other functional effects, but those effects remain poorly characterised in healthy individuals. Immune biomarker effects have been reported in some healthy adult trials, but biomarker changes are not the same as clinical outcomes, and evidence for hard endpoints in healthy people is limited.
This does not mean probiotic supplementation is without value in healthy individuals. There are specific contexts, antibiotic courses being the clearest example, where evidence supports use even in otherwise healthy people. The point is that choosing a probiotic on the basis of vague wellness positioning, without a specific evidence-matched indication, is unlikely to produce predictable results.
A note on fermented foods
Fermented foods including yogurt, kefir, kimchi, sauerkraut, and miso contain live microorganisms, and there is reasonable evidence that regular consumption of fermented foods supports gut microbiota diversity and some aspects of immune function. These effects, however, are more characteristic of a dietary pattern than a strain-specific intervention. They reflect the cumulative influence of varied microbial exposure through diet rather than the targeted delivery of a characterised strain at a clinical dose.
The organisms in fermented foods are not typically the same strains studied in probiotic trials. Their numbers and viability vary with production methods, pasteurisation, and storage. They cannot be used as proxies for the specific strains and doses used in clinical research. Fermented food consumption and supplemental probiotic use occupy different parts of the evidence landscape and should not be conflated.
The decision in practice
For most people considering a probiotic supplement, the process should work as follows. Identify the specific health context, antibiotic use, a digestive condition, a paediatric indication, vaginal health, or another specific concern. Consult the evidence for strains studied in that context; the strain entries in this library cover the five most clinically characterised options. Identify products that contain those strains at the relevant doses and with the strain properly labelled. Prioritise products with expiration-guaranteed CFU counts and named third-party testing. Be sceptical of marketing claims that exceed what the trial evidence for the contained strains actually supports.
It is also worth acknowledging that even a correctly chosen, well-characterised strain does not guarantee a response. Individual variation in gut microbiota composition and immune status means that outcomes differ substantially between people, and this cannot be predicted reliably in advance. A reasonable assessment window for most digestive applications is two to four weeks at the recommended dose. If no benefit is apparent by that point, the case for continuing is weak. Switching to a different strain with evidence for the same indication is more logical than increasing dose or extending the same product indefinitely.
For the substantial number of people considering probiotics without a specific indication, the honest answer is that the evidence base for general wellness supplementation is limited. This does not make supplementation unreasonable, but it does mean that expectations should be calibrated accordingly, and that high-cost products making expansive claims are unlikely to be more effective than simpler, well-characterised options at a fraction of the price.
For detailed evidence ratings, trial summaries, and population-specific guidance, see the individual strain entries in the Evidentia library: Lactobacillus rhamnosus GG, Saccharomyces boulardii, Lactobacillus reuteri, Lactobacillus acidophilus, and Bifidobacterium lactis BB-12. For a conceptual overview of how probiotic evidence works, see the Probiotics hub.