Rybelsus and the weekly injection carry the same active molecule. Nothing about semaglutide changes when it is pressed into a tablet; what changes is how much of it survives the trip into the bloodstream, how reliably, and under what conditions. That single difference reorganizes the dose numbers, the morning routine and the size of the gap between one patient and the next. Which form a seller will quote you, and at what price, is a separate question handled in the format article.
Ninety-nine percent of the tablet never arrives
A peptide swallowed on its own is digested before it can be absorbed. The oral products get around that by co-formulating semaglutide with sodium N-[8-(2-hydroxybenzoyl) aminocaprylate], abbreviated SNAC. Imaging and clinical work showed that absorption happens in the stomach rather than the intestine, is confined to a small area close to the surface of the dissolving tablet, requires SNAC to occur at all, and proceeds through the cells rather than between them, with no evidence of any effect on tight junctions.[1]
The efficiency of that route is the headline number. The prescribing information gives absolute bioavailability of 0.4% to 1% for Rybelsus 3, 7 and 14 mg and 1% to 2% for the Ozempic tablet strengths,[11] against 89% for subcutaneous semaglutide.[10] A population pharmacokinetic analysis pooling six clinical pharmacology trials put oral bioavailability at 0.8% under the recommended dosing conditions, rising with a longer post-dose fast and falling with a larger volume of water. Within-subject variability in that bioavailability was 137%, which once-daily dosing and a week-long half-life smooth down to 33% variability in steady-state exposure.[2]
Why the milligrams do not convert
The clearest demonstration that a milligram figure is not a dose equivalence sits inside one document. Rybelsus and Ozempic tablets are the same molecule from the same manufacturer under a single prescribing information, and that document states plainly that the two are not substitutable on a mg-to-mg basis. Its switching table pairs Rybelsus 7 mg with Ozempic tablets 4 mg, and Rybelsus 14 mg with 9 mg.[11] Two tablets of one drug, and the numbers on them differ by nearly a factor of two for the same effect.
Across routes the gap is far larger. The weight-management tablet is taken at 25 mg once daily, which is 175 mg swallowed over a week; the weight-management injection is 2.4 mg once weekly.[10] The label’s own bridge between routes is narrower still: a patient on the 0.5 mg weekly injection may switch, one week after the last shot, to 7 mg or 14 mg of Rybelsus daily.[11] No milligram arithmetic connects those figures, and the ladders that climb toward them are set out in the dose article.
Matched on the average, unmatched on the person
Exposure is the currency the milligrams are buying. At steady state, Rybelsus produces mean concentrations of 6.7 nmol/L at 7 mg and 14.6 nmol/L at 14 mg in type 2 diabetes.[11] The weight-management doses are far higher: the 25 mg tablet averages 77 nmol/L and the 2.4 mg weekly injection 75 nmol/L, in patients with obesity or overweight.[10] On the average, the two routes land within three percent of one another.
The same paragraph of the label carries the number that complicates it. Ninety percent of patients on the 25 mg tablet had average concentrations between 27 and 186 nmol/L. Ninety percent on the 2.4 mg injection sat between 51 and 110 nmol/L.[10] That is roughly a seven-fold span against a two-fold one. A patient at the low end of the tablet distribution is carrying about half the exposure of a patient at the low end of the injection distribution, and a patient at the high end is carrying more than the injection ever reaches. The label describes this as higher variability compared with subcutaneous administration, and it is the clinically important difference between the routes: the average is the same, and the person is not.
The morning rules, and which one actually matters
Every oral semaglutide product is taken on an empty stomach in the morning, with no more than four ounces of water, swallowed whole, with a wait of at least 30 minutes before food, other drinks or any other oral medication. A missed dose is skipped rather than doubled.[11] Those instructions are usually repeated as a list of equally weighted rules, and the trials behind them do not weight them equally.
In a food-effect trial, 78 healthy subjects were randomized to dose after a meal, after an overnight fast continued four hours, or under reference conditions. The fed arm produced limited or no measurable semaglutide exposure, while every subject in the fasting arm had measurable exposure. A second trial randomized 161 healthy men across eight combinations of water volume and post-dose fasting time. Exposure was not different between 50 mL and 120 mL of water (p = 0.541 for area under the curve) but rose significantly with a longer post-dose fast (p < 0.001).[3] The waiting is doing the work. The four ounces is a ceiling rather than the mechanism, and eating too soon is the error that removes the dose entirely.
One further consequence follows from taking a drug that delays gastric emptying at the same hour as everything else in the cabinet. Total thyroxine exposure rose 33% when a single 600 mcg levothyroxine dose was given with oral semaglutide at steady state,[11] which is why the 30-minute rule covers other tablets and not only breakfast — the wider interaction picture is in the oral medication article.
What the head-to-head data actually shows
There is no head-to-head. A systematic review published in 2026, which screened to June 2025 and included 30 studies — 12 randomized trials, nine systematic reviews, five comparative or observational studies and four pharmacokinetic studies — states that direct comparative trials of oral against subcutaneous semaglutide are lacking and calls for them.[4] Everything below is a comparison across separate trials with separate participants.
The placebo-controlled results cluster tightly. Oral semaglutide 50 mg reached −15.1% of body weight at 68 weeks in OASIS 1.[5] In OASIS 2, 201 adults in Japan and South Korea, a quarter of them with type 2 diabetes, reached −14.3% against −1.3% on placebo, a treatment difference of 13.07 percentage points (95% CI, −15.61 to −10.52).[6] The weekly 2.4 mg injection reached −14.9% in STEP 1.[7] The 25 mg tablet now carried on the weight-management label has its own trial, examined in the OASIS 4 article.
The closest thing to a route experiment compares the tablet with a different injected molecule. PIONEER 4 randomized 711 adults with type 2 diabetes to oral semaglutide escalated to 14 mg, subcutaneous liraglutide escalated to 1.8 mg, or placebo. At 26 weeks glycated hemoglobin fell 1.2% on the tablet and 1.1% on the injection, a difference of −0.1 percentage points (95% CI, −0.3 to 0.0), and weight fell 4.4 kg against 3.1 kg (difference −1.2 kg; 95% CI, −1.9 to −0.6; p = 0.0003).[8] A swallowed peptide outperformed an injected one. What it does not show is anything about semaglutide against itself.
Where the network puts the two routes, and where it splits them
A network meta-analysis published in 2026 pooled 262 trials and 99,791 participants, searched through November 2025, and placed both forms among the six agents carrying moderate-to-high certainty at one year. Oral semaglutide came out at −10.9% against lifestyle modification alone (95% CI, −12.7 to −9.1) and subcutaneous semaglutide at −9.8% (95% CI, −10.6 to −9.1).[9] On the pooled weight estimate, with the intervals overlapping heavily, the tablet is nominally ahead.
Two findings in the same analysis point the other way. Oral semaglutide appears in the group of agents with the highest discontinuation due to adverse events, with risk ratios across that group running from 1.9 to 4.2, and among those with the largest increase in gastrointestinal events; subcutaneous semaglutide appears in neither list. And subcutaneous semaglutide was the only drug in the whole network associated with reduced all-cause mortality (risk ratio 0.81; 95% CI, 0.72 to 0.93) and reduced myocardial infarction (0.72; 95% CI, 0.61 to 0.85), estimates the authors attribute largely to cardiovascular outcome trials in high-risk populations.[9] So the route that edges ahead on pooled weight is the one people leave more often, and the route behind it is the one carrying the mortality signal. The oral cardiovascular evidence is real but was collected in a different population, and is set out in the SOUL article.
The labels do not cover the same patients
One route has a broader license than the other. The injection is indicated for adults and for patients aged 12 and older with obesity, and carries an accelerated approval for noncirrhotic metabolic dysfunction-associated steatohepatitis with stage F2 to F3 fibrosis. The weight-management tablet is indicated for adults only, and for cardiovascular risk reduction and weight reduction — the adolescent and liver indications are not on it.[10] Same molecule, same company, narrower label, because the trials supporting those indications were run with a needle. The brand-by-brand version of that distinction is in the Wegovy and Ozempic article.
What this comparison settles
On potency, the two routes deliver comparable average exposure and comparable placebo-controlled weight results, in trials that were never run against each other. On reliability, the injection wins clearly: 89% bioavailability and a two-fold spread between patients, against roughly 1% and a seven-fold spread. On daily burden, the trade is one injection a week against a fasted half hour every morning, with the penalty for getting it wrong being an absent dose rather than a smaller one. On evidence for outcomes beyond weight, the two routes are not interchangeable and their labels say so.
Every figure here describes an FDA-approved product taken at a labeled dose. Compounded semaglutide is not FDA-approved, and the FDA does not review it for safety, efficacy or quality before it is dispensed; a compounded oral or sublingual preparation has no published bioavailability of its own, so none of the exposure figures above can be assumed to describe it. How the numbers on this site are established before publication is set out in the methodology.