Shedding is one of the few adverse events on this drug class that a person notices in a mirror rather than in a bathroom, and it is the one most likely to end a prescription that is otherwise working. It was recorded in the pivotal trials at rates several times placebo. It is also the textbook consequence of losing a large amount of weight quickly by any means at all, which is precisely what these drugs are sold to do. Both statements are true, and which one carries the weight decides whether a bigger percentage lost should be read as a reason to expect more of it.
The randomized rates, with both arms
SURMOUNT-1 is the clearest single record. Alopecia emerged during treatment in 32 of 630 participants on tirzepatide 5 mg (5.1%), 31 of 636 on 10 mg (4.9%) and 36 of 630 on 15 mg (5.7%), against 6 of 643 on placebo (0.9%).[1] SURMOUNT-3 recorded 20 of 287 (7%) against 4 of 292 (1.4%).[1] Pooled across the tirzepatide weight-loss program, alopecia appears in 5.4% (119 of 2,183) of treated participants against 0.9% (22 of 2,221) on placebo.[1]
A 2026 meta-analysis of nine interventional studies covering 4,114 GLP-1 users put the pooled risk ratio for hair loss at 3.252 (95% CI, 1.437 to 7.358). Restricted to randomized trials in overweight or obesity it rose to 3.587 (95% CI, 2.100 to 6.124), and the single-arm event rate was 3.9%.[2] A separate benefit–harm model drawing on eight randomized trials and 8,847 participants translated that into absolute terms: an excess of 57 alopecia events per 1,000 people treated over two years, with a wide interval of 10 to 176.[3] So the rate is neither trivial nor matched by placebo, which is where a reader arriving from the general side-effect record might reasonably have expected it to sit.
Which semaglutide is being asked about
The semaglutide numbers separate by dose and route rather than collapsing into one figure. Oral semaglutide at 50 mg daily recorded alopecia in 23 of 334 participants (7%) against 9 of 333 (3%) on placebo. The 2.4 mg weekly injection sat at roughly 3% against 1% across product trials of 2,116 people, and one clinic series found 1 case in 407 users at that dose (0.2%).[1] Doses below 2 mg weekly are rarely implicated at all.[1]
That is a dose gradient in name only. The 50 mg oral formulation and the 2.4 mg injection are the ones prescribed to produce large weight loss, and the diabetes doses below them are not. The exposure that tracks the shedding may be the pound count rather than the milligram count, and the rest of the evidence keeps pointing the same way.
The split by sex is larger than the split by drug
In the tirzepatide product trials, alopecia was reported by 7.1% of women against 0.5% of men.[1] A fourteen-fold gap between sexes inside a single program dwarfs every difference between molecules discussed above, and the mechanism behind it is not established.[1]
The surgical literature shows the same asymmetry. In 50 patients followed for six months after sleeve gastrectomy, hair loss occurred in 46% of the women and 10% of the men.[10] A meta-analysis of 18 bariatric studies found it more common in women, though its estimate was imprecise and did not reach significance: odds ratio 3.87 (95% CI, 0.59 to 17.59; p = 0.08).[8]Sex differences on these drugs are otherwise scarce, which is the point of the article on men and women; this is the exception.
What telogen effluvium is, and when it arrives
The pattern being described is almost always telogen effluvium: a diffuse shedding that follows a stressor rather than a bald patch that spreads from one. Acute telogen effluvium is defined as shedding lasting under six months, and the hair loss typically begins two to three months after the trigger. Around 95% of acute cases remit.[7]
The recognized triggers are a list this drug class sits squarely inside. Severe protein, fatty acid and zinc deficiency, chronic starvation and caloric restriction all appear on it, and essential fatty acid deficiency produces shedding two to four months after intake falls short.[7] Surgical trauma and chronic systemic illness are on the same list.[7] Nothing about that list requires a receptor agonist to be involved.
The comparison group lost the weight without a drug
Bariatric surgery has been producing rapid weight loss for decades, and the hair literature there is far larger. A 2025 meta-analysis of 41 articles and 7,044 patients put the incidence of hair loss after metabolic and bariatric surgery at 47%.[9] An earlier meta-analysis of 18 studies and 2,538 patients estimated 57% (95% CI, 42 to 71%), falling with longer follow-up.[8] A prospective series six months after sleeve gastrectomy found 56%.[10]
The surgical timing splits in two. Acute onset arrives within the first three months and is characteristically telogen effluvium; a later, chronic form appears around six months out and is the one attributed to nutritional deficiency.[11] Procedure matters in a way that supports the second reading: Roux-en-Y gastric bypass carried a higher rate than sleeve gastrectomy (OR 1.91) and a lower rate than duodenal switch (OR 0.41), ranking the three by how malabsorptive they are.[9] What a GLP-1 does and does not reproduce after one of these operations is a separate question, taken up in the post-surgical article.
The one study that put the two exposures side by side
A retrospective cohort in the TriNetX network compared 218,147 GLP-1 users with obesity and without type 2 diabetes against 143,471 controls. Tirzepatide was associated with increased telogen effluvium risk against other weight-loss medications, at a hazard ratio of 2.65. Against bariatric surgery — a comparator losing comparable weight with no incretin involved — the hazard ratio was 1.61 and not significant. No increased risk was found for semaglutide or liraglutide at all.[6][1]
Read alongside the trials, that is the shape of the answer. Risk is roughly flat across tirzepatide 5, 10 and 15 mg, so the dose is not the driver.[1] Risk rises with the weight actually lost: 5.3% among participants losing more than 20% of body weight against 2.5% among those losing less.[1] And the excess over other drugs collapses when the comparator is an intervention producing the same loss.
What does not exist is a study designed to separate the two. No randomized trial has isolated a follicular effect of the molecule from the effect of the weight it removes, and the systematic reviews say so directly, calling for prospective randomized work to establish causality and timing.[1][4] The bariatric comparison is observational and confounded by who chooses surgery. It is the strongest evidence available and it is not proof.
Spontaneous reports rank the drugs the other way around
Pharmacovigilance databases record what people and clinicians volunteer, and they produce a different league table. Across ten studies covering 626,894 GLP-1 patients for whom hair outcomes were reported, 9,933 (1.6%) had hair loss and 52 (0.008%) reported regrowth. Reporting odds ratios ran 1.24 to 2.46 for semaglutide, 0.83 to 1.73 for tirzepatide and 0.61 to 1.53 for liraglutide.[4] A 2026 analysis of 1,276 alopecia-related reports found semaglutide significantly associated after adjustment, at an odds ratio of 1.23 (95% CI, 1.11 to 1.35), while tirzepatide and dulaglutide showed lower or non-significant signals.[5]
So the randomized data make tirzepatide look worst and the spontaneous reports make semaglutide look worst. Reporting volume follows brand recognition and prescription counts, not incidence, and two of those three reporting odds ratio ranges cross 1.0 in the first place. Neither instrument is the truth on its own: trial capture undercounts a symptom nobody is asked about, and internet-era self-report overcounts the drug that is being talked about.
Protein, iron and zinc are the tractable part
On a much smaller plate, intake of everything falls, and three of those nutrients carry their own hair literature. In the bariatric meta-analysis, hair loss was associated with lower serum zinc (standardized mean difference −1.13; 95% CI, −2.27 to 0.01; p = 0.05), lower folic acid (SMD −0.88; 95% CI, −1.29 to −0.46; p < 0.0001) and lower ferritin (SMD −0.22; 95% CI, −0.38 to −0.05; p = 0.01) — but not with serum iron or vitamin B12.[8] What a trial actually prescribed people to eat is the subject of the diet article.
A normal laboratory result does not settle it either. Among 42 women followed after sleeve gastrectomy, 16 (41%) reported hair loss; zinc and iron were both significantly associated with it, yet mean levels in the affected patients sat inside the normal range, and only 3 (7.7%) had frankly low zinc.[12] A combined zinc-plus-iron value below 115 predicted shedding with 88% sensitivity and 84% specificity, at an odds ratio of 4 (p = 0.006).[12] On the drugs themselves, a 2026 review of micronutrient risk during incretin therapy concludes that most reported abnormalities are subclinical or indirect, and that prospective studies are still needed to define incidence — while flagging prior bariatric surgery, poor baseline diet and prolonged nausea as the conditions that raise the stakes.[13]
What a seller owes a reader here
Every rate above was measured on an FDA-approved product in a controlled trial. Compounded semaglutide and tirzepatide are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, so no alopecia incidence has ever been established for a compounded vial. The trial figures are the closest available estimate and they are a borrowing, which is worth holding in mind when comparing what plans include across the tirzepatide board.
Two conclusions do not follow from any of this. That acute telogen effluvium usually remits is a statement about 95% of cases over months, not a description of what the months feel like, and a person losing density in their thirties is not comforted by a population figure. And “it is the weight loss, not the drug” is an explanation of mechanism, not a reason to leave shedding off a side-effect list — the weight loss is the product being sold. A service that names it, says when it typically starts and says that it usually resolves has told the truth. One that omits it because the trials call the signal non-significant has used a statistical word to do a commercial job, and how any figure on this site is established is set out in the methodology.