3 providers are running promotional pricing right now. Two of them don't say when it ends. · Verified September 15, 2026
Evidence reviewed September 15, 2026
There is no clinical or regulatory definition of it. No prescribing label contains the word. No trial has ever tested it. And the professional society for obesity medicine has now said in print exactly what it is.
From the journal of the Obesity Medicine Association, published November 2025:
“Microdosing is a marketing term used to describe the administration of very small, often subtherapeutic amounts of obesity medications — at levels far below the Food and Drug Administration approved and commercially available doses.”
The same statement draws the distinction that sellers collapse. A doctor lowering your dose because you cannot tolerate it, or settling you at a lower approved dose, is ordinary medicine. Campaigns that start everybody on a non-approved dose are, in the statement's own words, “a commercial tactic rather than an evidence-based, clinical approach.”
The word appears nowhere in the prescribing information for Wegovy, Ozempic or Zepbound. The approved schedules all start at a dose intended only to get you started:
FDA label Zepbound, verbatim: “The 2.5 mg dosage is for treatment initiation and is not approved as a maintenance dosage.” Ozempic's 0.25 mg is initiation-only in the same way. No approved product exists below 0.25 mg of semaglutide or 2.5 mg of tirzepatide.
Part of what makes this hard to argue about is that “microdosing” is sold to mean at least four things that are not alike:
Unsupported No prospective trial of GLP-1 microdosing has ever been completed. Not for weight, not for longevity, not for inflammation, not for anything. The published literature consists of letters, commentaries, review articles and one preprint that has not been peer reviewed.
Randomised trial The one genuinely sub-starting dose ever randomised produced almost nothing. In a phase 2 trial, tirzepatide at 1 mg — below the 2.5 mg approved start — produced 0.9 kg of weight loss against 0.4 kg on placebo over six months. Half a kilogram, over half a year. At approved doses in the same class of trial: 5 mg lost 4.8 kg, 10 mg lost 8.7 kg, 15 mg lost 11.3 kg.
The trial sellers cite is not a microdosing trial, and the reason is arithmetic. The famous semaglutide dose-ranging study found that even its lowest arm lost 6.0% of body weight — which sounds like proof that tiny doses work.
That arm was 0.05 mg per day. Per week, that is about 0.35 milligrams — more than the 0.25 mg weekly dose the label calls a starting dose. The lowest arm of the trial is not a microdose. It is a larger dose than the one you begin on, given to people with a BMI of 30 or more, and it still produced less than half the effect of the top arm.
This is the part that gets quoted in microdosing marketing, and it is worth being precise about, because the findings themselves are excellent.
Randomised trial Heart attacks and strokes. In 17,604 people with existing cardiovascular disease and no diabetes, semaglutide cut major cardiovascular events from 8.0% to 6.5% (HR 0.80, 95% CI 0.72–0.90). The dose was 2.4 mg — the full maintenance dose.
Randomised trial Kidney disease. In 3,533 people with type 2 diabetes and chronic kidney disease, the trial was stopped early for benefit: kidney failure and death events fell by roughly a quarter (HR 0.76, 95% CI 0.66–0.88). The dose was 1.0 mg. This finding is now printed on the label.
Randomised trial Liver disease. In people with confirmed steatohepatitis, 2.4 mg resolved the condition in 62.9% versus 34.3% on placebo.
Every one of those trials used a full approved dose, in people with a diagnosed disease, alongside roughly 9–15% weight loss. Not one tested a fraction of the dose. Not one enrolled a healthy person — every one had a minimum BMI to get in.
There is no randomised trial of any of these drugs in adults of normal weight without diabetes, for any outcome at all. Researchers studying them for non-weight conditions routinely exclude lean people, as a safety matter.
Randomised trial Brain health and neuroprotection: falsified twice, at scale. In 3,808 people with Alzheimer's disease, oral semaglutide over two years produced no difference whatever in cognitive decline (2.3 versus 2.3 points, and 2.2 versus 2.1 in the companion trial). The open-label extension was discontinued. In Parkinson's disease, a GLP-1 drug was numerically worse than placebo at 96 weeks. These were disease populations, so they do not prove nothing happens in a healthy brain. They do mean neuroprotection is a hypothesis two large trials have failed to support — and it is still being sold as a benefit.
Unsupported Longevity. No human trial has ever measured lifespan on these drugs. The mouse study behind the headlines used female mice of a single strain. The human mortality figure people quote comes from a trial where it was a formally untested secondary result — and a later analysis found the gap was driven substantially by COVID-19 deaths.
Unsupported “Cravings without the weight loss.” This has never been tested. No study has given a sub-therapeutic dose and measured appetite with weight held constant. The term “food noise” only acquired a validated way to measure it in 2025, so every claim made before then was measuring something nobody had defined.
Preclinical Autoimmune conditions. Psoriasis has one open-label study of 31 people who also had diabetes. Hidradenitis suppurativa has a single 20-person trial of unknown status. Rheumatoid arthritis has no trial at all — cell work and one database review. A multiple sclerosis trial reports around 2030.
FDA safety communication Patients have administered five to twenty times their intended dose. FDA documented the mechanism: people were given U-100 insulin syringes and told to draw a number of “units”, and injected 50 units instead of 5. Prescribers made the same error converting milligrams to units, causing five- to ten-fold overdoses. Reported consequences included severe vomiting, severe hypoglycaemia, fainting, dehydration, acute pancreatitis and gallstones, with some hospitalisations. This matters here more than anywhere else, because drawing a small volume by hand from a vial is exactly what microdosing requires. The obesity medicine statement notes it “typically does not involve administration by a pen device” — the pen being the thing that makes a dose hard to get wrong.
Observational The reporting bears it out. As of 31 May 2026, FDA held 990 adverse-event reports for compounded semaglutide and more than 730 for compounded tirzepatide — and notes these are undercounted, because state-licensed pharmacies are not required to file them. US poison centres logged 5,713 cases between 2017 and 2022, four-fifths of them dosing errors, rising 81% in a single year.
These drugs are approved for a BMI of 30, or 27 with a related condition. One large microdosing programme lowered its own eligibility threshold to a BMI of 21 — inside the range the CDC calls healthy. Another screens specifically for people interested in “longevity, not weight loss.” A third advertises a target of five to twenty pounds.
Microdosing is the dose at which a drug approved for a disease can be sold to people who do not have it.
There is a second reason, and it is legal rather than clinical. Since the shortages ended in 2025, a compounding pharmacy may not make what is essentially a copy of an approved drug — and for a while the argument ran that a dose which is not a labelled dose is not a copy of a labelled dose.
Federal Register That argument has now been examined by the FDA on the record, and rejected in detail. In a May 2026 notice proposing to bar these drugs from compounding entirely, the agency took the microdosing case apart by name. It noted that the companies making it provided no reference supporting their claim about patients needing lower doses; never explained why the approved 0.25 mg presentation or simply a slower titration would not serve; could not identify a single specific alternative dose; and, asked for evidence that microdosing was clinically needed, cited a newspaper article. The FDA's reply, in substance: a general statement that a different dose might be “more suitable” does not mean the approved dose is unsuitable. The agency also rejected cost and supply as grounds: shortages and price are not what “clinical need” means. Its 2018 guidance had already closed the arithmetic route — a strength you can reach by taking a fraction or a multiple of an approved dose is not a meaningfully different strength, and a lower price is explicitly not a sufficient reason.
Fifth Circuit And the litigation is over. Compounders sued the FDA over both the tirzepatide and semaglutide shortage determinations and lost at every stage; the Fifth Circuit affirmed the agency in both cases on 27 August 2026. Anything you read describing this as an open legal question was written before that.
The American Association of Clinical Endocrinology is the only US body on record about microdosing itself. Its president, in October 2025: “AACE does not support the routine use of microdosing GLP-1RA medications.” That is a president's statement rather than a balloted guideline, and we say so. The other societies have issued nothing on microdosing specifically — which we report as an absence, not as approval.
The clearest line comes from Dr W. Scott Butsch of the Cleveland Clinic, and it is worth keeping: “We can adjust the dose to the individual needs of each person. And that's not microdosing — that's just the art of practicing medicine.”
Dr Jody Dushay of Harvard: “Microdosing GLP-1s is not a thing. There are no legitimate long-term data to support it.” Dr Daniel Drucker, who co-discovered the biology these drugs work on: “there is no clinical trial data supporting microdosing.”
Randomised trial It comes back when you stop. In the extension of the main semaglutide trial, participants had lost 17.3% of their weight at week 68. One year after stopping, they had regained 11.6 points of it — roughly two-thirds — and the improvements in blood pressure, blood sugar and cholesterol reverted with it. Which is the quiet arithmetic of paying monthly for a dose too small to do much: not a cheaper result, but a smaller one you still have to keep buying.
Lower doses are not fake medicine — a doctor settling you at Wegovy 1.7 mg or Zepbound 5 mg is prescribing an approved dose, and the lowest approved tirzepatide dose still produced 15% weight loss in trials. That is not what this page is about.
What this page is about is a word with no clinical meaning, attached to doses below anything ever tested, sold largely to people who do not meet the criteria for the drug, requiring the one handling step that FDA has linked to twenty-fold overdoses.
Every real benefit these drugs have — the heart attacks prevented, the kidneys protected, the livers healed — was measured at a full dose in someone with a diagnosed illness. Whether a fraction of that dose does anything at all in a healthy person is not a question the evidence has answered. It is a question nobody has asked.
Reviewed September 15, 2026. Figures were read from the trial publication, FDA document or professional statement cited, and each claim is graded for the kind of evidence behind it. Where a statement comes from a society president rather than a formal guideline, or from a preprint rather than a peer-reviewed paper, we say so on the claim. Several figures circulating in coverage of this topic were excluded because we could not verify them against a primary source.
We keep a table of every provider we track, ranked by what twelve months actually costs rather than the price they advertise.
That spread moves. 3 of these providers are running promotional pricing right now, and two of them will not say when it ends. We check every week. Want us to tell you when yours changes?
Your email and nothing else. Not your weight, your dose, or which page you were reading. We deliberately do not store that. No trackers, never sold.