Peptide results and timelines.
The most searched question in this field is some version of "how long until it works." The honest answer depends entirely on whether anyone has measured it. For the GLP-1 receptor agonists there are large randomised trials that map the whole curve week by week. For most research peptides there is nothing comparable, and the timelines circulating online are reconstructed from personal anecdote, which is the least reliable evidence there is. This guide explains what the trial data actually shows, why transformation content misleads, and how to tell a biomarker moving apart from an outcome changing.
Reviewed June 1, 2026
- Before and after content is the weakest form of evidence in this field. It has no control group, no denominator, no blinding, and a selection filter that only publishes the successes.
- In STEP 1, mean weight change at week 68 was -14.9% with semaglutide 2.4 mg versus -2.4% with placebo. In SURMOUNT-1, week 72 change was -20.9% at the 15 mg tirzepatide dose versus -3.1% with placebo. Both curves are gradual and still descending near the end.
- Healing peptide timelines are effectively unknowable from personal experience, because most soft tissue injuries improve on their own schedule whether or not anything is taken.
- A biomarker moving is not an outcome changing. IGF-1 rising tells you a drug reached its receptor, not that anything you care about improved.
- Withdrawal data is the clearest signal of all. One year after stopping semaglutide in the STEP 1 extension, participants had regained about two thirds of their prior weight loss.
Why before and after content is the weakest evidence in the field.
The single most common way people estimate a timeline is by watching someone else's transformation post and counting the weeks in the caption. This is understandable, and it is close to worthless as evidence. Every structural feature that makes a randomised trial informative is absent from a transformation post, and several features that make results misleading are present by design.
The problem is not that the person is lying. Most of them are not. The problem is that a single uncontrolled observation cannot distinguish between a drug effect, a placebo effect, a simultaneous change in diet or training, a natural fluctuation, and a photograph taken with better lighting and a different posture. All five produce the same picture.
- 01No control group. You cannot see what would have happened to the same person over the same weeks without the compound. Every trial in this guide exists specifically to answer that question, and in every one of them the placebo arm also changed.
- 02No denominator. You see the people it worked for. You do not see the people who took the same thing, felt nothing, and quietly stopped posting. This is survivorship bias, and in a field with no registry it is total.
- 03Confounded by everything else. People start compounds during periods of high motivation. They usually change food, sleep, training, and alcohol at the same time. Attributing the whole change to the vial is a choice, not a finding.
- 04No blinding. The person knows what they took, and expectation reliably shifts self-reported outcomes like pain, energy, mood, and sleep quality. These are precisely the outcomes most often reported for research peptides.
- 05Photographic variance. Lighting, camera angle, time of day, hydration, glycogen, pump, tan, and posture can produce a visible difference between two photos taken an hour apart with nothing in between.
- 06Selection of the endpoint after the fact. The before photo is chosen retrospectively, usually the least flattering one available. That is a different thing from a prespecified baseline measurement.
What the GLP-1 trials show about the shape of the curve.
The GLP-1 receptor agonists are the rare case where the timeline question has a real answer, because the pivotal trials ran long enough to plot the entire curve and published the week by week means. STEP 1 randomised 1,961 adults with obesity or overweight without diabetes to 68 weeks of once weekly subcutaneous semaglutide 2.4 mg or placebo, both alongside lifestyle intervention. Mean change in body weight from baseline to week 68 was -14.9% with semaglutide versus -2.4% with placebo, an estimated treatment difference of -12.4 percentage points. In absolute terms that was -15.3 kg versus -2.6 kg.
SURMOUNT-1 randomised participants with obesity to 72 weeks of once weekly tirzepatide or placebo. Mean percentage weight change at week 72 was -15.0% at 5 mg, -19.5% at 10 mg, and -20.9% at 15 mg, against -3.1% with placebo. Half of the 10 mg group and 57% of the 15 mg group lost 20% or more of body weight, compared with 3% on placebo.
- 01The first weeks are titration, not results. Both programmes escalate the dose over roughly the first four months. Meaningful separation from placebo appears early, but the steepest part of the curve arrives after the target dose is reached.
- 02The rate is fastest in the middle, not the beginning. The curve is not linear. Weekly rate of loss increases as dose escalates, then gradually flattens.
- 03It had not fully plateaued at the end of the trial. Both curves were still descending at 68 and 72 weeks. That is one reason the trials are quoted in weeks rather than months: the endpoint is over a year out.
- 04Placebo arms lost weight too. -2.4% in STEP 1 and -3.1% in SURMOUNT-1, both with lifestyle intervention. That difference is the part an uncontrolled anecdote can never separate out.
- 05The response distribution is wide. In STEP 1, 86.4% of the semaglutide group lost at least 5% and 50.5% lost at least 15%. Those are not the same people. A single reported number is a mean, and roughly half the participants fell below it.
Titration weeks are not results weeks.
A large share of the disappointment reported in the first two months comes from a category error. The approved labels start well below the studied maintenance dose and escalate on a fixed schedule specifically to reduce gastrointestinal adverse reactions. The Wegovy injection label starts at 0.25 mg once weekly for four weeks and reaches the 2.4 mg maintenance dosage at week 17. The Zepbound label starts at 2.5 mg once weekly for four weeks, a dosage the label explicitly states is for treatment initiation and is not approved as a maintenance dosage.
So a person four weeks in is not four weeks into the treatment that was studied. They are four weeks into a ramp designed for tolerability. Comparing that experience against a headline number from a 68 week trial is comparing two different things. The same logic applies in reverse: side effects reported during escalation are also not a stable picture of what maintenance feels like, which is why the trials describe gastrointestinal events as typically transient and occurring primarily during dose escalation.
Why healing peptide timelines are unknowable from personal experience.
The peptides most often discussed for tendon, ligament, gut, and soft tissue repair sit in a completely different evidence position from the GLP-1 agonists. Most have no published randomised human trials at all. That absence is not a technicality when the outcome in question is injury recovery, because injuries have a natural history: they get better on their own, on a timeline that varies enormously between people and between tissues.
This is the core reason personal experience cannot resolve the question. If a strained tendon would have settled in eight weeks regardless, and someone starts a compound in week three and feels better by week eight, the compound gets the credit for something that was already happening. There is no way for that individual to know the difference, because they only ran the experiment once and there was no control arm.
The size of this effect has been measured directly. A meta-analysis of three-armed trials that included a no-treatment arm alongside placebo and active treatment found substantial spontaneous improvement in the untreated groups, particularly for subjective and continuous outcomes. Separately, the Cochrane review of placebo interventions across all clinical conditions found that placebo effects on patient-reported continuous outcomes such as pain are real but modest, and that much of what is casually called a placebo response is in fact the natural course of the condition plus regression to the mean.
- 01Injuries resolve on their own schedule. Most soft tissue complaints improve over weeks to months with no intervention. Any compound started during that window will appear to work.
- 02Symptoms fluctuate. Pain and stiffness vary day to day. People notice the good days more after starting something new.
- 03Behaviour changes at the same time. Starting a recovery protocol usually coincides with rest, load management, physiotherapy, or sleep changes. All of those have real effects.
- 04The outcome is self-reported. Pain, mobility, and “feeling better” are exactly the outcome types most sensitive to expectation and least suited to unblinded self-assessment.
- 05There is no registry. No one collects the cases where nothing happened, so the observed “success rate” online has no denominator.
A biomarker changing is not an outcome changing.
This distinction does more work than any other idea in this guide. A biomarker is something measured in blood or on a scan. An outcome is something a person experiences: injury healed, function restored, event avoided, life longer. The formal definitions were set out by the Biomarkers Definitions Working Group, which distinguished biomarkers from clinical endpoints and set a high bar for when a biomarker may stand in as a surrogate for an outcome. That bar is rarely met.
The clearest example in this field is IGF-1. Growth hormone secretagogues reliably raise IGF-1, and the CJC-1295 study in healthy adults documented prolonged elevation of GH and IGF-1 after single doses. That is a real pharmacological finding, and it demonstrates that the compound reached its target and produced a downstream signal. It is not evidence that tendon healing accelerated, that body composition improved in a durable way, or that anything a person would notice changed. Those would be separate outcomes requiring separate trials.
The same applies to almost every number people track. A shift in a lipid panel, a change in fasting insulin, a rise in a growth factor, or a movement in an inflammatory marker each tells you the body responded. Whether that response translates into an outcome is a different question, answered by a different study design, and the history of medicine contains many surrogates that moved in the right direction while the outcome did not follow.
- 01Would I notice this without a blood test? If the only evidence of an effect is a lab value, the effect is currently a biomarker finding.
- 02Has the biomarker been validated as a surrogate for this specific outcome? Validation is outcome-specific. A marker can be a good surrogate for one endpoint and useless for another.
- 03Did anyone measure the outcome directly? If a trial could have measured the thing you care about and instead reported the biomarker, that choice is informative.
Regression to the mean, and why people start at their worst point.
Regression to the mean is a statistical fact, not a theory. When a measurement varies over time and you select a moment when it is extreme, the next measurement will on average be closer to that person's typical value, purely because the extreme included a component of random variation. No intervention is required for this to happen.
This matters enormously here because of when people start. Almost nobody begins a compound on an average day. They begin during the worst flare of shoulder pain in two years, at their heaviest weight, after the worst sleep month they can remember, or immediately after a bloodwork result that came back unusually poor. Selecting the extreme is exactly the condition under which regression to the mean produces the largest apparent improvement.
A recent primer for evidence-based medicine practitioners lays out the mechanics and the practical consequence: any single-arm before and after comparison that begins at a selected extreme will overstate the effect, and the more extreme the selection the larger the overstatement. This is one of the main reasons uncontrolled clinical series routinely report benefits that vanish when a control arm is added.
What the post-cessation data shows.
Withdrawal designs are among the most informative studies in this area, because stopping a drug and watching what happens is a much cleaner test of whether the drug was doing the work than starting it and watching what happens. The GLP-1 programmes ran several of these, and the results are consistent.
The STEP 1 trial extension followed 327 participants for a year after all treatment, including the lifestyle intervention, was discontinued at week 68. Mean weight loss to week 68 had been 17.3% with semaglutide and 2.0% with placebo. By week 120, the semaglutide group had regained 11.6 percentage points of lost weight and the placebo group 1.9 percentage points, leaving net losses from baseline of 5.6% and 0.1%. Cardiometabolic improvements seen at week 68 reverted toward baseline for most variables. The authors concluded that participants regained about two thirds of their prior weight loss and that ongoing treatment appears to be required to maintain the improvements.
STEP 4 approached the same question from the other side. After a 20 week run-in on semaglutide during which participants lost a mean of 10.6%, they were randomised either to continue or to switch to placebo. From week 20 to week 68, mean body weight change was -7.9% with continued semaglutide versus +6.9% with placebo. SURMOUNT-4 repeated the design with tirzepatide: after a 36 week lead-in producing a mean 20.9% reduction, participants continuing tirzepatide changed -5.5% from week 36 to week 88 while those switched to placebo regained 14.0%.
- 01The effect is maintenance-dependent, not curative. Weight regain begins promptly after discontinuation in every one of these designs.
- 02The metabolic improvements track the weight. Blood pressure, waist circumference, and lipid changes largely reverted alongside the regain in the STEP 1 extension.
- 03Regain is not total, at least within a year. Net loss from baseline remained positive at week 120 in the STEP 1 extension, though much smaller.
- 04Continued treatment kept producing further loss. In both STEP 4 and SURMOUNT-4 the continuing arms were still losing during the maintenance period, which is another sign the curves had not plateaued.
- 05This reframes the timeline question entirely. “How long until it works” has an answer. “How long do I need to keep taking it” is the question the withdrawal data actually addresses, and the trial answer so far is indefinitely.
A more honest way to think about your own timeline.
None of this means personal observation is useless. It means personal observation answers a narrower question than people usually ask of it. It can tell you whether something is tolerable, whether you are willing to keep doing it, and whether a specific measurable thing has moved. It cannot tell you causation, and it cannot generate a timeline that generalises to anyone else.
- 01Fix the measurement before you start. Decide what you will measure and how, and record a real baseline over more than one day, so you are not comparing against a selected extreme.
- 02Change one thing at a time. Starting a compound, a new training block, and a new diet in the same week guarantees an uninterpretable result.
- 03Track objectively where possible. A repeated measurement under standardised conditions is far more informative than a recollection of how a week felt.
- 04Expect the first weeks to be uninformative. For anything with a long half-life or a titration schedule, early weeks reflect the ramp, not the treatment.
- 05Write down your prediction in advance. Recording what you expect before you start is the cheapest available protection against reinterpreting an ambiguous result as a success.
- 06Treat a strong early result with more suspicion, not less. Dramatic short-term change in a symptom that fluctuates is the signature of regression to the mean.
FAQ.
How long do GLP-1 medications take to work?
The pivotal trials ran 68 weeks (STEP 1, semaglutide) and 72 weeks (SURMOUNT-1, tirzepatide), and the weight curves were still descending at the end. Separation from placebo appears within the first months, but the labeled dose escalation schedules mean the studied maintenance dose is not reached until roughly week 17 for Wegovy injection. Early weeks reflect titration rather than the treatment that was studied.
Why are before and after photos considered weak evidence?
They have no control group, no denominator (you never see the people it did not work for), no blinding, and they are usually confounded by simultaneous diet, training, and sleep changes. The before photo is also chosen retrospectively. Every trial in this area exists precisely because uncontrolled observation cannot separate a drug effect from natural change, expectation, and regression to the mean.
How long does BPC-157 take to work for an injury?
There is no published randomised human trial establishing a timeline, so any specific number of weeks is an estimate from anecdote or from rodent models. The deeper problem is that most soft tissue injuries improve on their own over weeks to months, so an individual cannot distinguish a compound effect from natural healing without a control arm. Meta-analyses of three-armed trials show substantial spontaneous improvement in untreated groups.
My IGF-1 went up. Does that mean it is working?
It means the compound reached its target and produced a measurable downstream signal. That is a biomarker finding, not an outcome. Whether raised IGF-1 translates into healing, body composition change, or anything you would notice is a separate question requiring a trial that measures those outcomes directly. Biomarkers only substitute for outcomes when they have been formally validated for that specific endpoint, which is uncommon.
Will I regain the weight if I stop a GLP-1?
The trial data consistently shows substantial regain. In the STEP 1 extension, one year after stopping semaglutide and the lifestyle intervention, participants had regained about two thirds of their prior weight loss and most cardiometabolic improvements had reverted toward baseline. In SURMOUNT-4, participants switched to placebo regained 14.0% from week 36 to week 88 while those continuing tirzepatide lost a further 5.5%.
Why did I lose weight faster at month four than month one?
The labeled schedules escalate the dose over roughly the first four months to reduce gastrointestinal adverse reactions, so the first weeks are spent below the studied maintenance dose. The trial curves are not linear: the rate of loss increases as the dose escalates and then gradually flattens. This is expected behaviour, not an unusual response.
What is regression to the mean and why does it matter here?
When a measurement fluctuates and you select an extreme moment, the next measurement tends to be closer to typical purely by chance, with no intervention involved. People almost always start compounds at their worst point: peak pain, heaviest weight, worst bloodwork. That is exactly the condition that produces the largest apparent improvement, which is why uncontrolled before and after comparisons systematically overstate effects.
Everyone in the forum reports the same timeline. Doesn't that count for something?
Consistency across anecdotes mostly reflects shared expectations rather than shared biology. Forums have no denominator: people who felt nothing usually stop posting, so the visible sample is filtered toward responders. Shared narratives about when something 'kicks in' also shape when people report noticing it. Consistent anecdote is not the same evidence category as a controlled trial.
How much weight did people lose in the placebo groups?
In STEP 1 the placebo group lost a mean 2.4% at week 68, and in SURMOUNT-1 the placebo group lost 3.1% at week 72. Both placebo groups also received lifestyle intervention. That change is what an uncontrolled personal result cannot separate out from the drug effect, and it is a substantial fraction of what many people report anecdotally from unstudied compounds.
Is animal data enough to estimate a human timeline?
No. Rodent healing runs on a different clock and scale, injuries in animal models are deliberately standardised rather than the varied real-world presentations people have, and dose scaling between species is not a simple conversion. Animal data can justify running a human trial. It cannot substitute for the result of one.
Sources.
- [1]Wilding JPH et al.: Once-Weekly Semaglutide in Adults with Overweight or Obesity (STEP 1) · N Engl J Med, 2021 (PMID 33567185)
- [2]Wilding JPH et al.: Weight regain and cardiometabolic effects after withdrawal of semaglutide: the STEP 1 trial extension · Diabetes Obes Metab, 2022 (PMID 35441470)
- [3]Rubino D et al.: Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance (STEP 4) · JAMA, 2021 (PMID 33755728)
- [4]Jastreboff AM et al.: Tirzepatide Once Weekly for the Treatment of Obesity (SURMOUNT-1) · N Engl J Med, 2022 (PMID 35658024)
- [5]Aronne LJ et al.: Continued Treatment With Tirzepatide for Maintenance of Weight Reduction in Adults With Obesity (SURMOUNT-4) · JAMA, 2024 (PMID 38078870)
- [6]Spontaneous improvement in randomised clinical trials: meta-analysis of three-armed trials comparing no treatment, placebo and active intervention · BMC Med Res Methodol, 2009 (PMID 19123933)
- [7]Hrobjartsson A, Gotzsche PC: Placebo interventions for all clinical conditions · Cochrane Database Syst Rev, 2010 (PMID 20091554)
- [8]Regression to the mean: a primer for evidence-based medicine practitioners · BMJ Evid Based Med, 2025 (PMID 41390172)
- [9]Biomarkers Definitions Working Group: Biomarkers and surrogate endpoints: preferred definitions and conceptual framework · Clin Pharmacol Ther, 2001 (PMID 11240971)
- [10]Teichman SL et al.: Prolonged stimulation of GH and IGF-I secretion by CJC-1295 in healthy adults · J Clin Endocrinol Metab, 2006 (PMID 16352683)
- [11]WEGOVY (semaglutide) prescribing information: dosage escalation schedule and clinical studies · FDA / DailyMed
- [12]ZEPBOUND (tirzepatide) prescribing information: dose escalation schedule and maintenance dosage · FDA / DailyMed
Cite this page
PepCue. “Peptide results and timelines.” PepCue, reviewed June 1, 2026. https://www.pepcue.app/guides/results-and-timelines.
Track your protocol in PepCue.
Log doses, run the calculators, follow the evidence, and keep peptides, GLP-1s, TRT and labs in one private timeline.