GLP-1 microdosing: what the dose-ranging trials actually show

What microdosing means here, and what it does not

Microdosing a GLP-1 means taking a weekly dose below the lowest amount the manufacturer sells. There is no agreed definition, no approved schedule, and no trial that has yet reported on the practice as a strategy. The 0.25 mg semaglutide starting step on the label is a titration rung, not a microdose.

GLP-1 stands for glucagon-like peptide-1, a hormone your gut releases after a meal that tells the pancreas to make insulin and tells the brain you have eaten enough. Semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound) are long-acting copies of that signal, engineered to last days rather than the natural hormone's few minutes. Both are available as weekly injections, started low and stepped up over months, a process called titration, so the body can adapt to the nausea.

"Microdosing" is the internet's word for deliberately stopping short of that climb, or dropping below its first rung. There is no medical definition of it, no approved schedule that contains it, and no dose the label describes that way [5]. That vagueness matters, because two very different questions get filed under the same word.

The first is a starting question: can somebody who has never taken one of these get a worthwhile result from a dose far below the approved range, and skip most of the side effects? The second is a maintenance question: once the weight is off, can it be held on a smaller dose instead of stopping or staying at full strength? The evidence for those two is in completely different states, and most articles blur them together.

The consensus answer you will find almost everywhere is that no studies exist. That is not quite right, and the part that is wrong is the useful part. Two microdosing trials are registered and enrolling, and neither has posted a result [15] [16]. But four randomized trials have given these drugs at doses far below the ones used for weight management, because that is what a dose-ranging study does, and a fifth, published in May 2026, tested cutting an established dose down. Almost nobody quotes any of them. This page is those five trials, with every dose figure treated as a study parameter, not a plan for a reader.

The dose-ranging trial almost nobody quotes

A randomized trial tested semaglutide at daily doses far below anything sold today. It gave 957 adults with obesity daily injections for 52 weeks. Mean weight change was 2.3 percent on placebo and 6.0 percent on the smallest arm, rising step by step to 13.8 percent on the largest.

Before semaglutide had an approved dose for weight management, somebody had to find out what dose to approve. That study was published in The Lancet in 2018. It enrolled 957 adults with a body mass index of 30 or above and no diabetes, randomized them across eight countries, and ran for 52 weeks [1]. For this topic, the doses it tested were 0.05 mg, 0.1 mg, 0.2 mg, 0.3 mg and 0.4 mg, alongside a liraglutide comparator and a matching placebo [1].

Those are microdose-sized numbers. The smallest arm, 0.05 mg, is a fifth of the 0.25 mg step the semaglutide label now starts on, and roughly a fiftieth of the 2.4 mg dose used in the phase 3 obesity trial [5].

The result is a clean dose-response curve. At week 52, estimated mean weight loss was 2.3 percent on placebo, then 6.0 percent at 0.05 mg, 8.6 percent at 0.1 mg, 11.6 percent at 0.2 mg, 11.2 percent at 0.3 mg and 13.8 percent at 0.4 mg [1]. Every semaglutide arm beat placebo by a statistically significant margin, including the smallest. Weight loss of 10 percent or more occurred in 10 percent of the placebo group against 37 to 65 percent of those on 0.1 mg or above [1].

Read on its own, that looks like the strongest argument for microdosing anyone has ever published: the smallest dose tested still produced roughly two and a half times the weight loss of placebo, and the curve rises steeply through the low end rather than sitting flat until some threshold. It deserves to be in the conversation. It is also where almost everyone who quotes a low dose stops reading, and the next line of the methods section changes what the number means.

Why a weekly microdose is not that trial’s low dose

That trial injected daily, not weekly. Semaglutide’s half-life in healthy adults is roughly 145 to 168 hours, which is why the marketed product is weekly. Taking 0.05 mg once a week delivers about a seventh of what the trial’s smallest arm received across the same seven days.

The methods line that matters is short: all treatment doses were delivered once daily [1]. The 2018 trial predates the once-weekly obesity formulation. Its 0.05 mg arm took 0.05 mg every day [1], which is 0.35 mg over a week. Its 0.4 mg arm took 2.8 mg a week, comfortably above the 2.4 mg used in the phase 3 program [5].

That single fact rescales the whole curve. A half-life is the time it takes the body to clear half of a drug, and a systematic review of semaglutide pharmacokinetics puts the subcutaneous half-life at roughly 145 to 168 hours in healthy adults, about a week [2]. That long tail is why the marketed product is injected weekly: last week's dose is still there when this week's arrives, so the level in the blood climbs over the first month and then plateaus. What the body responds to is total exposure, not one injection's size.

Line the two up on that basis and the comparison collapses. A 0.05 mg weekly dose delivers about one seventh of the drug that the 2018 trial's smallest arm delivered over the same week, because that arm injected the same amount seven times [1]. The lowest weekly figure that would match the trial's smallest arm is 0.35 mg, which is above the label's 0.25 mg starting step [5] rather than below it. The weekly trials, which the next section covers, go lower than that, and they are the ones worth reading for a weekly dose.

There is a second correction in the same part of the protocol. The arms all started at 0.05 mg per day and climbed from there to their assigned target, escalating every four weeks in the main schedule [1], so the larger groups spent months on the way up. Only the 0.05 mg group was at its target from day one [1]. That makes the smallest arm the cleanest read of a sustained low dose in the whole study, which is a point in its favor, and it still does not make it a weekly microdose.

The lowest weekly doses anyone has actually tested

Weekly dosing has been studied down to 0.1 mg of semaglutide, and at that dose almost nothing happened. Blood sugar fell 0.6 percent against 0.5 percent on placebo over 12 weeks. Tirzepatide at 1 mg weekly moved body weight by 0.9 kg against 0.4 kg on placebo over 26 weeks.

Weekly dosing has its own dose-finding study, and it speaks most directly to a weekly microdose. In 2016 a phase 2 trial randomized 415 adults with type 2 diabetes to once-weekly semaglutide, and the low arms were given 0.1 mg, 0.2 mg, 0.4 mg or 0.8 mg with no dose escalation at all, held flat for 12 weeks [13]. That is the shape of a microdosing protocol, run as a randomized trial.

Its results posting gives the change in HbA1c, a blood test reflecting average glucose over the previous three months, for each arm. At 0.1 mg weekly it fell 0.6 percent, against 0.5 percent on placebo, rising to 0.9 percent at 0.2 mg, 1.0 percent at 0.4 mg and 1.4 percent at 0.8 mg [14]. Body weight fell by up to 4.8 kg, at the highest escalated arm [13]. On that evidence the sponsor took 0.5 mg and 1.0 mg weekly with a four-week escalation forward into phase 3 [13], and the arms at the bottom went no further.

Those phase 3 doses are where the weight data live. SUSTAIN 1 randomized 388 adults with type 2 diabetes to weekly semaglutide at 0.5 mg or 1.0 mg or to placebo for 30 weeks, and mean body weight fell 3.73 kg on 0.5 mg and 4.53 kg on 1.0 mg against 0.98 kg on placebo, from a 91.93 kg baseline [3]. That is a real effect, and about a quarter of what 2.4 mg weekly produced in the phase 3 obesity trial [5].

Tirzepatide's low end is starker. Its phase 2 trial randomized 318 adults with type 2 diabetes to weekly tirzepatide at 1 mg, 5 mg, 10 mg or 15 mg, to dulaglutide, or to placebo for 26 weeks [4]. On blood sugar it worked: HbA1c fell 1.06 percent at 1 mg against 0.06 percent on placebo [4].

On weight it did not. Across the four tirzepatide arms mean body weight change ranged from 0.9 kg to 11.3 kg, against 0.4 kg on placebo [4], and the trial's results posting gives the per-arm figures: 0.9 kg at 1 mg, 4.8 kg at 5 mg, 8.7 kg at 10 mg and 11.3 kg at 15 mg [11]. That gap, half a kilogram against placebo, is not a weight-loss effect anyone would notice. The glucose benefit and the weight benefit come apart at the low end, which matters if weight is the point.

The side-effect argument, checked against the numbers

Lowering the dose does reduce nausea, but it does not remove it. At semaglutide 0.5 mg weekly, one in five participants reported nausea against roughly one in twelve on placebo. One reading of the tirzepatide data hints that the wanted effect fades faster across the dose range than the unwanted one, though that signal is indirect.

The most persuasive case for microdosing is not about weight. It is that these drugs make a lot of people feel unwell, that the gastrointestinal side effects are clearly dose-related, and that a smaller dose should therefore buy most of the benefit for a fraction of the misery. The first two parts are well supported. The third is where the trial data push back.

In SUSTAIN 1, nausea was reported by 20 percent of the 0.5 mg group against 8 percent on placebo, with diarrhea in 13 percent against 2 percent [3]. That is the lowest weekly semaglutide dose ever taken to phase 3, and it still made roughly one participant in five nauseated, around two and a half times the placebo rate. Gastrointestinal side effects were also the main reason people left the trial in every arm, placebo included.

The tirzepatide phase 2 data let you watch both curves at once, because it recorded appetite suppression as an adverse event alongside the gastrointestinal ones. At 1 mg, gastrointestinal events were reported by 23.1 percent against 9.8 percent on placebo, while decreased appetite was reported by 3.8 percent against 2.0 percent [4]. Step up to 5 mg and gastrointestinal events reach 32.7 percent while decreased appetite jumps to 20.0 percent [4].

Read across those two doses and the ratio moves the wrong way for the microdosing argument. Gastrointestinal events roughly doubled over placebo at the lowest dose while the appetite signal barely moved; at 5 mg the appetite signal multiplied five-fold while gastrointestinal events grew by about half again [4]. The effect people are chasing appears to fade faster as the dose drops than the effect they are trying to avoid. A patient-reported adverse event is a rough proxy for therapeutic appetite suppression rather than a measure of it, so this is a signal and not a settled finding. It is also one of the few places in the published record where the two can be watched together, and it does not say what the microdosing pitch says.

Stepping down after reaching a goal is a different question

This version of the question got its first real answer in May 2026. SURMOUNT-MAINTAIN randomized 378 adults who had already lost weight on tirzepatide to continue, to drop to 5 mg, or to switch to placebo. The reduced-dose group held far more of the loss than placebo and less than continuing.

A lot of people who say they are microdosing are not starting low at all. They have finished a full titration, reached a weight they are happy with, and want to come down rather than stay at full strength or stop. Until recently the honest answer was that nobody had tested it: STEP 4 and SURMOUNT-4, the two trials usually cited here, each compared a full dose against placebo with no reduced-dose arm in between [6] [7].

That changed on 12 May 2026, when The Lancet published SURMOUNT-MAINTAIN. After a 60-week open-label period in which participants lost weight on tirzepatide at the maximum dose they could tolerate, 378 adults were randomized to continue that dose, to reduce to 5 mg, or to switch to placebo for a further 52 weeks. At week 112, model-based mean weight change from baseline was 21.9 percent with the maximum tolerated dose, 16.6 percent with the reduced 5 mg dose and 9.9 percent with placebo, with all comparisons significant [10].

The rescue figures make the same point in plainer terms. Rescue treatment was offered to anyone regaining at least half of what they had lost, and it was needed by 8 percent of those who continued, 25 percent of those who dropped to 5 mg and 67 percent of those who switched to placebo [10]. The trial's authors conclude that reducing to 5 mg might provide a valuable alternative to discontinuation, while noting that individual responses vary.

Two things follow, and they pull in opposite directions. A reduced maintenance dose is no longer speculative: one large trial now supports it as a middle option after weight has been lost. But the trial's reduced arm was 5 mg [10], an ordinary dose inside the labeled range rather than a microdose, so what it supports is stepping down under supervision, not the much smaller amounts this article is about. Our guide to what the tirzepatide withdrawal trials found covers the discontinuation data in more depth.

Where a microdose comes from, and why that matters

An approved multi-dose pen can be microdosed by counting its unnumbered clicks. A Diabetes Care commentary puts a semaglutide pen at about 72 clicks and tells clinicians how to teach the count, while saying plainly that the practice is off-label and that no trial has validated it.

This part of the topic is about the device, not the pharmacology, and it is where most articles go wrong. A semaglutide multi-dose pen has a dosing window and delivers roughly 72 unnumbered clicks, and a 2025 Diabetes Care commentary tells clinicians to counsel patients to count them for an individualized dose, handout included [17]. A fraction of a labeled step needs no compounding at all.

The same commentary is clear about the status: microdosing a pen is off-label, endorsed by no manufacturer and validated by no trial, and it suits only patients with the literacy, vision and dexterity to follow a count [17]. It names tirzepatide vials as the other route [17], and vials are where the handling risk concentrates. Compounded means prepared by a pharmacy rather than manufactured and tested as an approved product, and the concentration is whatever that pharmacy chose.

A regional poison control center shows where that goes wrong. Three patients were reported after incorrect self-administration of compounded semaglutide obtained from compounding pharmacies and an aesthetic spa. Two of the three had self-administered ten-fold dosing errors, all three had notable nausea, vomiting and abdominal pain lasting days, and one patient reported dosing in milliliters and units rather than in milligrams [8]. The authors are explicit about the mechanism: vials do not carry the safety features of a prefilled pen, and syringes not intended for semaglutide invite confusion between three different units.

The smaller the intended dose, the less a mistake looks like one, and one of those three needed an anti-sickness drug and intravenous fluids [8].

The supply picture has changed underneath the trend. The FDA declared the tirzepatide shortage resolved in December 2024 and the semaglutide shortage resolved in February 2025, which removed the legal basis for pharmacies to compound copies of drugs that are commercially available [9]. In a notice published on 1 May 2026 the agency proposed not to include semaglutide, tirzepatide or liraglutide on the list of bulk substances that outsourcing facilities may compound from [12]. Our explainer on the 2026 FDA peptide reclassification covers the wider shift.

What would actually settle this

Starting low is still the half with no reported trial. Settling it needs a trial that randomizes people to a defined low weekly dose against a standard titration and reports weight, side effects and how many stay on treatment for a year. Until one reports, the practice is unproven rather than disproven, and the difference matters.

Pull the five trials together and the position is narrower than either side of the argument admits. Semaglutide has a real dose-response curve that stays significant at very small daily amounts [1], so the idea that low doses do nothing is wrong. Those amounts were once daily [1], and the half-life is close to a week [2], so the weekly doses people actually inject are several times smaller. Weekly dosing has been randomized down to 0.1 mg, and at that dose blood sugar barely separated from placebo [13] [14]; the lowest weekly doses carried into phase 3 produced modest weight loss with real nausea [3], and at the bottom of the tirzepatide range the weight effect all but vanished while the side effects did not [4]. Coming down after a full course now has evidence behind it, at an ordinary approved dose rather than a small one [10].

What would resolve the starting question is not complicated in design, only in cost: randomize people to a defined low weekly dose against a standard titration, run for at least a year, and report weight change, adverse events and how many in each group were still taking the drug at the end. That last endpoint is the one worth watching, because the strongest case for a lower dose is not that it works as well, it is that more people might tolerate it long enough to find out, and none of the trials above set out to measure it.

Until a trial of that shape reports, the accurate description is that starting low is unproven rather than disproven, and the two are not the same. Nobody, including this page, can tell you what a low dose would do for you, and anyone selling a microdosing chart is selling something the literature does not contain. If you are weighing these compounds against each other, our GLP-1 comparison tool sets each drug's trial results out side by side, and the semaglutide mastery course works through the STEP, SUSTAIN and SELECT programs in detail. The questions worth taking to a prescriber are what a given dose was actually shown to do, what the alternatives are if side effects are the obstacle, and what the plan is for holding a result once it is reached.