Six current prescribing documents in this class say something about fluid, and between them they contain one number. It is not a target. It is a four-ounce maximum, it belongs to the oral tablet, and it exists so the tablet is absorbed rather than so the patient is hydrated. Everything else — every warning about volume depletion, every counseling line, every sentence in every Medication Guide — is written without a quantity. The kidney warning those sentences sit under is covered in the dehydration article; this page is about the other half, which is what goes in.
The census, with its control attached
Read on September 15, 2026 against the current labels for Wegovy, Ozempic, Zepbound, Mounjaro, Saxenda and the semaglutide tablet, the Patient Counseling Information section of all six contains the same sentence: advise patients to take precautions to avoid fluid depletion.[1][2][3] The Medication Guide of all six contains the same second sentence, addressed to the patient: it is important for you to drink fluids to help reduce your chance of dehydration.[1][2][3] Neither sentence names a volume, a frequency, a beverage or a way to tell.
The word hydration itself, standing on its own rather than inside dehydration, appears zero times in any of the six. That finding needs a control, because a search that reads nothing returns the same answer as a search that finds nothing. The same six documents contain the phrase “gastric emptying” between six and eight times each, and “oral medication” between three and nine times each. These labels are not quiet about what the drug does to absorption. They are quiet about drinking.
The one number is a ceiling, and its own label could not detect it
The semaglutide tablet is the exception, and it points the opposite way from what a reader expects. Its Dosage and Administration section instructs taking one tablet orally once daily on an empty stomach in the morning with water (up to 4 ounces of water), and not with any other liquid, followed by a wait of at least 30 minutes before eating food, drinking beverages or taking other oral medications.[3]
The Clinical Pharmacology section of the same document then undercuts the precision of that limit. In a ten-day study, healthy subjects took semaglutide tablets with 50 mL or 120 mL of water under fasting conditions across post-dose fasting periods of 15, 30, 60 or 120 minutes. Absorption was higher after a longer post-dose fast, and there were no clinically significant differences in semaglutide absorption between the two water volumes.[3] The variable that mattered was the waiting, not the water. Anyone reading the four-ounce figure as a hydration instruction has it backward twice: it is a maximum rather than a minimum, and it governs a tablet rather than a person. The wider absorption problem it belongs to is in the oral medication article.
Where the liters everyone quotes came from
The figures behind most fluid advice are the Adequate Intakes for total water published in 2005: 3.7 L per day for men and 2.7 L per day for women aged 19 to 30. Drinking water and beverages supplied 3.0 L and 2.2 L of those totals, about 81%, with the remainder coming from food.[4]
How those numbers were derived is the part that almost never travels with them. The report states that normal hydration can be maintained over a wide range of water intakes, and that the Adequate Intake is therefore set based on the median total water intake from United States survey data, rounded to the nearest 0.1 L. It says plainly that a wide range of intakes is compatible with normal hydration and that the AI should not be interpreted as a specific requirement.[4] A figure defined as the population median cannot also be a threshold most of the population fails; by construction, half of them already exceed it.
The data underneath show why the committee wrote it that way. Serum osmolality is the primary indicator of hydration status, and across deciles of intake in adults aged 19 to 50 it barely moves: men in the first decile drank a mean of 1.69 L per day at a mean osmolality of 279 mOsm/kg, and men in the tenth decile drank 7.93 L at 280 mOsm/kg. For women the same comparison runs 1.25 L against 6.16 L, both at 277 mOsm/kg.[4] A near-fivefold difference in intake moved the measurement by one unit or none.
Why the advice still lands differently on this drug
The same report describes how people stay in balance day to day, and the description contains the reason this question comes up at all. Over a few hours a deficit can open from reduced intake or increased losses, but on a day-to-day basis fluid intake is driven by the combination of thirst and the consumption of beverages at meals.[4]
One of those two drivers is what this drug class is prescribed to reduce. A measured cut of roughly a third in intake at a single test meal — the figure, and what it does and does not mean, is in the eating article — is also a cut in the drinking that happens around a meal, in a mechanism whose author named it as half of daily fluid regulation. Fewer meals, smaller meals and shorter meals each take a share of it, and a reduced sensation of wanting things does not obviously spare thirst.
That is an argument, not a measurement. No trial in this class has reported fluid intake as an outcome, which means the size of the gap between the two mechanisms is unquantified. The losses at the other end — vomiting, diarrhea and the volume they carry — have been measured, and belong to the vomiting article. The sensation most likely to be mistaken for thirst is covered in the dry mouth article.
Drinking more has been randomized, and it did less than expected
The strongest test of “drink more water and the kidney does better” was run in the population where the stakes are highest. The CKD WIT trial randomized 631 adults with stage 3 chronic kidney disease at nine centers to coaching that increased water intake or coaching to maintain usual intake, for twelve months. Mean age was 65 and mean baseline eGFR was 43 mL/min/1.73 m².[5]
The intervention worked as an intervention. Mean 24-hour urine volume was 0.6 L per day higher in the hydration group (95% CI, 0.5 to 0.7; P < .001), and plasma copeptin, a marker of vasopressin activity, fell by 2.2 pmol/L (95% CI, −3.9 to −0.5; P = .01). It did not change the outcome it was built to change. Mean eGFR change was −2.2 mL/min/1.73 m² in the hydration group and −1.9 in the control group, an adjusted between-group difference of −0.3 (95% CI, −1.8 to 1.2; P = .74). Patient-reported quality of health differed by 0.2 points (P = .22). The authors noted the trial may have been underpowered to detect a clinically important difference.[5]
That result answers a chronic question, not an acute one. It says that in adults with established kidney disease, a year of drinking measurably more did not slow decline. It says nothing about someone who has kept nothing down for three days, which is the situation the labeled warning describes and a different physiological event entirely. The chronic kidney outcomes for this drug class are in the kidney article. What the trial does settle is that “more water is protective” is an assumption with a randomized test behind it that came back null.
What is in the glass is not neutral
Volume is only one variable, and a trial designed to separate the others found that the usual ranking is wrong. Seventy-two euhydrated men each drank 1 L of still water or one of twelve other beverages over 30 minutes, with urine collected for four hours, to produce a beverage hydration index. Four-hour urine output was lower than the still-water control of 1,337 g after an oral rehydration solution (1,038 g), full-fat milk (1,052 g) and skimmed milk (1,049 g), all P < 0.001. The two-hour index came out at 1.54 for the rehydration solution, 1.50 for full-fat milk and 1.58 for skimmed milk.[6]
The list of beverages that were not different from water is the useful half: cola, diet cola, hot tea, iced tea, coffee, lager, orange juice, sparkling water and a sports drink all produced cumulative four-hour urine output indistinguishable from still water.[6] Coffee did not dehydrate anyone in that design, which is the opposite of the folk rule; the caffeine question in this context is taken up in the caffeine article. The beverages that beat water contained electrolytes or energy, which is the mechanism: solute slows the diuretic response to a fluid load.
The limits are as real as the result. Those were young euhydrated men given a single 1 L bolus, not people who had been unable to eat for two days, and the index measures retention over hours rather than recovery from a deficit. Nothing in it is a recommendation for a specific drink.
There is a ceiling, and no label sets it
The 2005 report declined to set a Tolerable Upper Intake Level for water, because healthy people excrete an excess readily. It records the exception anyway: acute water toxicity has been reported after rapid consumption of quantities that greatly exceeded the kidney’s maximal excretion rate of approximately 0.7 to 1.0 L per hour.[4] A person who has been told to drink more, feels unwell, and responds by drinking fast and eating nothing is taking in volume without solute, and that is the combination the exception describes.
Which is where this stops being a content question and becomes a clinical one. How much fluid, of what kind, and whether a dose should be held at all during an episode are judgments that depend on a medication list, a blood pressure and a renal function result. None of the six labels answers them, and a website cannot.
What a compounded prescription removes from this
Both sentences in the census above live in an FDA-reviewed document. Compounded semaglutide and tirzepatide are not FDA-approved, and the FDA does not review them for safety, efficacy or quality before they are dispensed. A compounded vial ships without prescribing information and usually without a Medication Guide, so the only two pieces of fluid guidance the class publishes — thin as they are — may never reach the person injecting it.
That makes the counseling a seller provides the whole of it. Whether a service tells a patient what volume depletion looks like, and whether it checks renal function at all before or during escalation, is a distinguishing fact rather than a formality, and it is recorded in each provider review.