Bariatric surgery or weight-loss medication: what the studies report

Bariatric surgery and weight-loss medication are two treatment categories that have been studied in very different ways. Surgery has decades of prospective follow-up. The current medications have large randomised trials, but only a few years of long-term data. This guide lays out what each body of evidence actually reports, and where the two cannot be compared directly.

Nothing here is a ranking. The studies were run in different populations, with different designs and different endpoints, and they answer different questions. What follows is what the published papers state, with every number linked to its source in the evidence drawer at the bottom.

What each option is

Metabolic and bariatric surgery is a one-time operation that changes the size or the routing of the stomach and small intestine. The two procedures that appear most often in current research are sleeve gastrectomy and Roux-en-Y gastric bypass. One analysis of the 2023 MBSAQIP registry included 200,865 patients who had one of those two operations that year — 133,620 sleeve gastrectomies and 67,245 gastric bypasses. Older procedures — vertical banded gastroplasty and gastric banding — appear throughout the long-term literature because they were the operations available when those studies started recruiting in the late 1980s.

Weight-loss medication is an ongoing prescription. The two most-studied current options are Zepbound (tirzepatide), a weekly injection, and Wegovy (semaglutide), available as a weekly injection and as a daily tablet. Foundayo (orforglipron) is a daily pill approved in April 2026. All of them are labelled for use alongside a reduced-calorie diet and increased physical activity, and all of them are taken on an ongoing basis. Withdrawal trials exist for tirzepatide and semaglutide; orforglipron has not been studied in a withdrawal trial.

Weight loss reported in each set of studies

In clinical trials of the medications, participants lost the following on average. In SURMOUNT-1, 2,539 adults with obesity or overweight with weight-related conditions took tirzepatide for 72 weeks and lost an average of 20.9% of body weight at the 15 mg dose; about 89% lost at least 5% and half lost at least 20%. In STEP 1, 1,961 similar adults took semaglutide 2.4 mg for 68 weeks and lost an average of 14.9%; about 86% lost at least 5% and 69% lost at least 10%.

In the surgical cohorts, the Swedish Obese Subjects (SOS) study followed 2,010 people who had surgery alongside 2,037 matched people receiving usual care. Maximum weight reduction came 1 to 2 years after the operation: 32% with gastric bypass, 25% with vertical banded gastroplasty and 20% with banding. The control group’s average weight changed by less than 2% over the same period.

One study has attempted a direct comparison. Published in Obesity in 2026, it analysed 44,025 adults with a BMI of 35 or above who either started an injectable GLP-1 medication or had surgery between 2018 and 2024, weighting the groups to make them more comparable. At one year, average total weight loss was 5.4% with semaglutide, 9.1% with tirzepatide, 24.4% with sleeve gastrectomy and 29.8% with gastric bypass. At three years, the reported figures were 7.4% for semaglutide, 22.0% for sleeve gastrectomy and 28.4% for gastric bypass; the study reports no three-year figure for tirzepatide. The authors concluded that “bariatric surgery was associated with greater weight loss than GLP-1RAs among patients eligible for both options.”

Worth noting: the medication figures in that real-world study are well below the trial averages. This is an observational study of people using these drugs in ordinary care, where prescriptions lapse, doses are not always escalated to the maximum, and insurance coverage changes. It measures something different from a trial that supplies the drug and monitors adherence for 68 or 72 weeks.

Durability: what happens over years

This is where the two evidence bases diverge most sharply, because one has 20 years of follow-up and the other does not yet exist at that length.

In SOS, mean body weight change in the surgery group was −23% at 2 years, −17% at 10 years, −16% at 15 years and −18% at 20 years. The control group was at 0%, +1%, −1% and −1% at the same time points. So the surgical cohort did regain — from a peak of −23% to around −17% — and then held roughly there for the next two decades.

For the medications, the equivalent data is about what happens when treatment stops. In SURMOUNT-4, participants who had lost weight on tirzepatide for 36 weeks were randomised to continue or switch to placebo. Over the following 52 weeks, those who stopped regained an average of 14% of body weight, while those who continued lost a further 5.5%. That pattern is covered in detail in our guide on what happens when you stop and on the stopping medication page.

The honest limitation on this section: no medication has been followed for 20 years the way SOS followed its surgical cohort, so there is no like-for-like durability comparison to make.

What the studies report about risks

The risks reported in each literature are different in kind, not just in size. The surgical studies below report a perioperative window and long-term nutritional findings; the medication figures below are side effects recorded during ongoing treatment. Neither list is exhaustive, and each comes from a different study.

These two columns are not a like-for-like comparison. They come from different studies, measure different kinds of event, and use different denominators — a 30-day surgical complication rate and a trial side-effect rate cannot be read against each other.

Beyond weight, SOS reported outcomes the medication trials have not run long enough to report. Overall mortality in the surgery group had an adjusted hazard ratio of 0.71 against usual care. Type 2 diabetes developed at 6.8 cases per 1,000 person-years after surgery versus 28.4 in controls (adjusted hazard ratio 0.17). Diabetes remission was more common at 2 years (odds ratio 8.42) and at 10 years (odds ratio 3.45). In the 2020 life-expectancy analysis, after a median of about 24 years, 22.8% of the surgery group and 26.4% of controls had died; adjusted median life expectancy was 3.0 years longer in the surgery group than in controls, and 5.5 years shorter than in the general population.

Eligibility, as the guidelines and labels state it

The two sets of criteria are written in different units, which makes them awkward to line up. The 2022 ASMBS/IFSO statement uses numeric BMI thresholds. It states that surgery “is recommended for individuals with BMI >35 kg/m², regardless of presence, absence, or severity of co-morbidities”; that it “should be considered in individuals with BMI of 30–34.9 kg/m² who do not achieve substantial or durable weight loss or co-morbidity improvement using nonsurgical methods”; and that it “is recommended in patients with T2D and BMI >30 kg/m²”. The same statement notes that “clinical obesity in the Asian population is recognized in individuals with BMI >25 kg/m²” and that “there is no upper patient-age limit to MBS”.

The current FDA labels for Zepbound and Wegovy do not state a numeric BMI cut-off in their indications sections. Both describe eligibility by category: use alongside a reduced-calorie diet and increased physical activity, in adults with obesity, or in adults with overweight in the presence of at least one weight-related comorbid condition. Insurers often apply their own numeric criteria on top of the label, which is a separate question from what the label says. Our eligibility page and the BMI calculator cover how those thresholds are applied in practice.

Cost, in one paragraph

One insurance-claims study has compared the two on money rather than kilograms. Using MarketScan claims for 6,748 insured patients with a BMI of 35 or above and type 2 diabetes, drawn from claims between 2016 and 2021, adjusted three-year out-of-pocket spending was $7,752 for semaglutide, $5,980 for sleeve gastrectomy and $6,591 for gastric bypass; total spending across the three groups was not statistically different. That is one commercially insured US population with diabetes, over three years, and it does not transfer to uninsured or self-paying situations. For what the medications themselves cost month to month — including self-pay prices and savings programmes — see the cost page and the comparison tool.

The two are not always alternatives

One randomised trial has tested using them together rather than choosing between them. BARI-STEP, a double-blind randomised placebo-controlled trial published in Nature Medicine in 2026, enrolled 70 adults who were at least a year past gastric bypass or sleeve gastrectomy and had lost less than 20% of their weight from the operation. Over 68 weeks, mean weight change was −18.0% with semaglutide 2.4 mg and +0.4% with placebo, an adjusted treatment difference of −19.18%. Adverse events were consistent with the known profile of semaglutide, with eight serious adverse events and no treatment-related deaths.

There is also a practical intersection in the other direction: anaesthesia guidance on holding GLP-1 medications before any operation, bariatric surgery included, is covered in our guide on GLP-1 drugs before surgery.

This is a decision for you and your doctors

Everything above is what the published studies report. None of it tells you which path fits your body, your medical history, your insurance, or your life — and this site is not in a position to. Surgery involves a surgeon, an anaesthetist and a dietitian; medication involves a prescriber and, usually, a pharmacy benefit. Both conversations are ones to have with clinicians who can see your records.

The questions that tend to matter in those appointments are specific ones: what the studies show for someone with your particular conditions, what follow-up each route requires, what your plan covers, and what happens if the first approach does not work out. Our talk to your doctor page has a list you can take in with you.