The short version
- In a 2006 trial of 362 women with IBS, Bifidobacterium infantis 35624 at 100 million CFU beat placebo on abdominal pain and bloating, while the same organism at 10 billion CFU did not.
- CFU counts how many live organisms are in a serving, not how much benefit they produce, and a total for a 12-strain blend tells you nothing about the dose of any single strain.
- A review of 228 randomised trials found probiotic efficacy tracks the specific strain and the specific condition: L. rhamnosus GG worked for paediatric antibiotic-associated diarrhoea but not for five other conditions tested.
- The strain designation — the letters and numbers after the species name, such as 35624 — is the only part of a probiotic label that can be traced back to a clinical trial.
- When 16 commercial probiotics containing bifidobacteria were checked by DNA-based methods, only one matched its bifidobacterial label claims perfectly across all samples tested.
In 2006, Whorwell and colleagues ran a multicentre trial in 362 women with irritable bowel syndrome. Everyone got the same organism — Bifidobacterium infantis 35624 — at one of three doses — 1 × 10⁶, 1 × 10⁸ or 1 × 10¹⁰ CFU — or placebo, for four weeks.
The middle dose won. At 1 × 10⁸ CFU the probiotic beat placebo on abdominal pain, on the composite symptom score, and on bloating, bowel dysfunction, incomplete evacuation, straining and gas. The other two doses — including the one that was a hundred times larger — were not significantly different from placebo.
Hold that next to the front of almost any probiotic bottle you have ever picked up, where the biggest thing on the label is a number of billions.
What CFU actually measures
CFU means colony forming units: a count of how many viable organisms are in a serving, established by culturing them and counting what grows. It is a real measurement. It is just an inventory number, not an outcome number.
Two things to check on the label before you take the count seriously. First, when the count is guaranteed — at time of manufacture, or through the printed expiry date. Those are very different promises, because live organisms die in storage. Second, whether the count is given per strain or only as a total for the blend. A "50 billion CFU, 12 strains" label tells you nothing about how much of any one of them you are getting.
The number that carries the evidence is the strain
Probiotics are identified at three levels: genus, species, and strain. Bifidobacterium is the genus. longum is the species. 35624 is the strain. That last alphanumeric tag looks like a serial number, and functionally it is one — it identifies the specific organism that was actually put through the trials.
This is not pedantry. McFarland, Evans and Goldstein reviewed 228 randomised controlled trials covering 25 probiotic products — 15 single-strain and 10 multi-strain mixtures — specifically to test whether efficacy tracks strain and condition. It does, sharply. For preventing antibiotic-associated diarrhoea, L. casei DN-114001 and L. reuteri 55730 were effective while L. rhamnosus GG was not. L. rhamnosus GG, one of the most-studied probiotics in the world, showed efficacy for paediatric antibiotic-associated diarrhoea but not for five other conditions tested.
Their warning about the literature is worth repeating, because it explains why probiotic headlines are so contradictory: reviews and meta-analyses "often fail to report strain designations and many inappropriately pool together different strains or species of probiotics into the same subgroup." Pool a strain that works with four that don't, and you get a null result that means nothing.
"Adequate amounts" is strain-specific, not universal
The definition everyone works from — reaffirmed by the ISAPP expert panel in 2014 — is "live microorganisms which when administered in adequate amounts confer a health benefit on the host." The load-bearing phrase is adequate amounts, and the Whorwell trial is exactly why. Adequate is whatever dose that strain was shown to work at, for that outcome. There is no universal threshold, and there is no evidence that stacking more CFU on top of an effective dose does more.
Does it survive the trip?
This is the other question worth asking, and the honest answer is that it depends entirely on the organism and the formulation, and that most labels won't tell you. Delayed-release capsules, enteric coating and spore-forming organisms are all approaches to the same problem. What you can do without any special knowledge is check whether the potency is guaranteed through expiry, follow the storage instructions on the bottle, and treat a product that makes no claim about either as one that hasn't thought about it.
And whether it is even the organism on the label
Lewis and colleagues took 16 commercial probiotic products containing bifidobacteria and checked, by DNA-based methods confirmed with culture, whether the contents matched the label. Only one of the 16 matched its bifidobacterial label claims perfectly across all samples tested. They also found variation pill-to-pill and lot-to-lot within the same product.
Given how much of probiotic efficacy is strain-level, mislabelling at the subspecies level is not a rounding error. It means the product in your hand may not be the product in the trial.
What we'd actually tell you
Turn the bottle over and look for strain designations. Named strains with letters and numbers after the species mean somebody can point you to a trial. A blend of species with no strain tags means nobody can, whatever the CFU count says.
Twenty billion CFU of identified, evidenced strains is a better purchase than a hundred billion of anonymous ones. Give it four weeks — the Whorwell trial's own window — and judge it on one specific thing you can actually notice, not on a general sense of wellness.
And rule out the boring causes first. Persistent changes in bowel habit, blood, unintended weight loss or pain that wakes you up are a doctor's appointment, not a supplement decision. We would much rather you got that checked.
Good questions
Is a probiotic with more CFU better?
No. In a 2006 trial of 362 women with IBS, Bifidobacterium infantis 35624 at 100 million CFU beat placebo on abdominal pain, bloating and bowel dysfunction, while the same organism at 10 billion CFU did not. CFU is an inventory number, not an outcome number. Adequate dose is whatever that specific strain was shown to work at.
What do the numbers after the bacteria name on a probiotic label mean?
They are the strain designation, and they are the part of the label that can be traced to a clinical trial. Probiotics are identified by genus, species and strain — Bifidobacterium longum 35624, for example. A review of 228 randomised trials found efficacy tracks the specific strain, so a blend listing only species names cannot point you to any evidence.
How long should I stay on a probiotic before deciding it did nothing?
Four weeks is a reasonable window, because that is the length of the trial that produced the strain-and-dose result described here. Judge it against one specific thing you can actually notice, such as bloating after a particular meal, rather than a general sense of wellness. If nothing has changed by then, stop. That is information, and it saves you a subscription.
Am I wasting my money on a 50 billion CFU, 12-strain blend?
Possibly, and the label may not let you find out. A total CFU count across 12 strains tells you nothing about how much of any single organism you are getting, and efficacy in the trial literature is strain-specific. Twenty billion CFU of named, evidenced strains is a better purchase than a hundred billion of anonymous ones, whatever the front of the bottle says.
Do probiotics need to be refrigerated?
Check the bottle, because live organisms die in storage and the right handling differs by formulation. The more useful question is when the CFU count is guaranteed: at time of manufacture, or through the printed expiry date. Those are very different promises. A product that makes no claim about either, and gives no storage instruction, has not thought about it.
Sources
- Whorwell PJ, et al. Efficacy of an encapsulated probiotic Bifidobacterium infantis 35624 in women with irritable bowel syndrome. Am J Gastroenterol, 2006. View study
- McFarland LV, Evans CT, Goldstein EJC. Strain-Specificity and Disease-Specificity of Probiotic Efficacy: A Systematic Review and Meta-Analysis. Front Med, 2018. View study
- Hill C, et al. The International Scientific Association for Probiotics and Prebiotics consensus statement on the scope and appropriate use of the term probiotic. Nat Rev Gastroenterol Hepatol, 2014. View study
- Lewis ZT, et al. Validating bifidobacterial species and subspecies identity in commercial probiotic products. Pediatr Res, 2016. View study
These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.
