Most of what moves the scale from one morning to the next is water and the food and drink still in your gut, not body fat. A change in body fat builds slowly, so it shows up in the trend over several weeks rather than in any single reading.
Why can the scale jump overnight?
A scale weighs everything at once: muscle, bone, fat, water, and whatever you have recently eaten and drunk. Fat changes slowly; water and gut contents can change noticeably within a day. The common “3,500-kcal rule” counts a pound of weight as 3,500 kcal, about 7,700 kcal per kilogram (Thomas et al., 2013). The rule is too simple for predicting weight loss (see below), but it shows why an overnight jump of a kilogram cannot be mostly new fat: that would take thousands of kilocalories beyond what you used that day.
What moves your weight from day to day?
Glycogen and the water stored with it. The body keeps carbohydrate as glycogen in the liver and muscles, and glycogen is stored with about three to four parts water (Kreitzman et al., 1992). A classic 1970 study reported that total body water varies with muscle glycogen (Olsson and Saltin, 1970). The 1992 paper, a short report built on 11 people measured over four days of a very low-calorie diet, discusses how this water-laden glycogen moves weight up and down early in a diet, and why regain looks exaggerated when a lot of carbohydrate is eaten again.
Salt. Salt affects how much water the body holds on to. In two long space-flight simulation studies (105 and 205 days), 10 healthy men ate controlled diets with 12, 9 or 6 g of salt a day, and more salt led the kidneys to conserve more water (Rakova et al., 2017). That study was not about everyday scale readings, and we found no good source that puts a number on how much a salty meal adds, so we do not give one.
Meal and fiber volume. Food and drink weigh something until they are digested and passed, so a large late dinner can still be on the scale the next morning. Fiber adds weight to stool: in a review of intervention trials in healthy people, cereal and vegetable fibers added more stool weight than fruit fibers, and less fermentable fibers added more than easily fermented ones (de Vries et al., 2016). None of that is fat.
Exercise. Sweating loses water, enough that sports scientists estimate a person’s sweat rate by weighing them before and after exercise (Sawka et al., 2007). A reading taken straight after a workout can come out low. Long or hard exercise also uses glycogen, and as muscles refill it, the water stored with it comes back too (Kreitzman et al., 1992), so the number can rise again in the following days without any change in body fat.
The menstrual cycle. In a study that weighed 42 women twice a week, body weight was about 0.45 kg higher during menstruation than in the comparison week, matched by a rise of about the same size in extracellular water (estimated by bioimpedance), with no other significant change in body composition (Kanellakis et al., 2023). The authors note that earlier studies, which used different methods, gave conflicting results, so one small study is not the final word.
The weekly rhythm. In a retrospective look at 80 adults and 4,657 weigh-ins, weight tended to be highest on Sunday and Monday and to fall through the week, and the authors suggest this weekend-to-weekday swing should be considered normal (Orsama et al., 2014).
How big are the normal swings?
Fewer good numbers exist than you might expect. The sourced ones are:
| What | Average size in the study | Who was studied |
|---|---|---|
| Within-week swing | 0.35% of body weight, about 0.25 kg at 70 kg | 1,421 adults in a European trial on keeping lost weight off (Turicchi et al., 2020) |
| Menstruation | about 0.45 kg, matched by extracellular water | 42 women (Kanellakis et al., 2023) |
| Christmas period | a mean rise of 1.35%, not fully reversed in the following months | 1,062 adults in the same 2020 trial |
These are group averages, not limits for one person. We found no systematic review that measures typical day-to-day scale noise, so we do not quote one. Several factors can stack on the same day, so a single reading can sit further from the trend than any one figure suggests. The Christmas result is a reminder that not every rise reverses on its own.
Why does the 3,500-kcal rule overpredict?
The rule assumes every 3,500 kcal of deficit removes a pound, at a steady rate, for as long as the deficit lasts. Thomas and colleagues compared its predictions with seven weight-loss experiments in which intake was supervised or objectively measured, and describe it as a rule that “grossly overestimates actual weight loss” (Thomas et al., 2013). The body is not a fixed account: as weight falls, the energy it uses falls too, so the same intake leaves a smaller gap over time (Hall et al., 2011).
Dynamic models include that adaptation. In Hall and colleagues’ model, body weight responds slowly to a lasting change in intake, with about half of the eventual change reached in around a year, and adults with more body fat are expected to lose more for the same cut in intake, and to take longer to settle (Hall et al., 2011). It is a mathematical model, not a trial, but it explains why a steady trend tends to slow over the months and why a straight-line forecast from the 3,500-kcal rule runs ahead of reality.
How do you read the trend?
- Similar conditions. Weighing at the same time of day, after the toilet, before eating, drinking or exercising, and in similar clothing reduces several of the sources of noise above.
- A smoothed line rather than single readings. A moving average, or a weighted average that counts recent days more, shows the direction under the noise. In smart-scale data from 50 people, Kalman smoothing and an exponentially weighted moving average matched observed weights best when filling in missing days (Turicchi et al., 2020). That study was about filling gaps, and no research we found names a best averaging window for everyday use, so any window is a choice, not a rule.
- Like with like. Because of the weekly rhythm, a weekly average, or one Monday against the next, is fairer than a Friday against a Monday.
- Weeks, not days. A direction that holds across several weekly averages says far more than any two readings.
What can a forecast say, and what can it not?
A weight forecast projects the recent trend forward. It can show the direction and rough pace if things carry on as before. It cannot know what you will eat next month, how much your expenditure will adapt, or whether this week’s readings are mostly water. Forecasts that start from logged food carry another uncertainty: the NIH notes that people often underestimate how much they eat (NIDDK Body Weight Planner). A forecast date is best read as an estimate that will move, most of all in the first weeks or when the trend is nearly flat.
IQFood shows the trend under the day-to-day noise in the weights you log, plus a forecast with a date for your goal, worked out from your recent weight trend and what you log. Like any forecast, that date is an estimate, not a promise.
Weighing is optional
In one six-month trial of 91 adults with overweight who were trying to lose weight, daily weighing did not appear to cause psychological harm: there were no significant differences in depressive symptoms, disordered-eating thoughts or binge eating, though the daily-weighing group did report more dietary restraint (Steinberg et al., 2014). Tracking is not right for everyone, though. In a survey of 493 college students, people who used calorie trackers reported more concern about eating and more dietary restraint, and the authors wrote that for some individuals these devices “might do more harm than good” (Simpson and Mazzeo, 2017). That was a cross-sectional study, so it cannot show cause.
Weighing less often, or not at all, is a reasonable choice. The NHS lists signs such as spending a lot of time worrying about weight and body shape, eating very little food or exercising too much, and advises anyone who thinks they may have an eating disorder to see a GP (a family doctor) as soon as they can (NHS, eating disorders).