Should You Eat Back Your Exercise Calories?
Short answer
Usually not all of them, and often none of them. Most calorie targets are built with an activity level already baked in, so when a watch hands you 520 kcal for a spin class and you add that on top, you have counted the same hour twice — and the second count is the one made up of guesses.
The question worth asking is not "how many did I burn?" but "does the number I am eating to already assume I train?"
How the double count happens
Standard targets come from an estimate of resting metabolism multiplied by an activity factor. Those factors run from roughly 1.2 for someone who barely moves to around 1.7 or higher for someone who trains most days and has a physical job. A person who ticks "moderately active" because they lift three times a week has already had those sessions priced into their target.
Then the watch reports the session separately, and it gets added again. Two 500 kcal sessions double-counted across a week is around 1,000 kcal — enough to swallow a large part of a modest weekly deficit. This is the single most common reason people describe training hard, eating to their number, and watching the scale sit still.
If you selected "sedentary" and genuinely are sedentary apart from planned exercise, the logic for eating some of it back is stronger. Most people who ask this question did not select sedentary.
Why the burn figure is generous
Three separate things inflate it.
Gross versus net. Your watch reports everything you burned during the hour, including the calories you would have burned anyway sitting on the sofa — roughly 60 to 80 kcal per hour for many adults. Only the extra above that is new. A reported 350 kcal run might be about 280 kcal of genuinely additional expenditure. The 45-minute yoga class showing 180 kcal is closer to 120.
Heart rate is a proxy, not a measurement. Estimating energy cost from heart rate works reasonably well for steady aerobic work — running at an even pace, cycling on the flat — because there the relationship between heart rate and oxygen consumption is fairly predictable. It works much worse for weight training, intervals, hot rooms, caffeine, stress, poor sleep and dehydration, all of which lift heart rate without a matching rise in energy cost. A heavy set of squats can spike your heart rate to 160 and cost remarkably little in calories.
Steps are an inference stacked on an inference. A step tracker detects a pattern of acceleration, decides it was a step, applies an assumed stride length, and converts distance to calories using your height and weight. Each stage carries error, and the errors compound. As a rough sanity check, walking costs on the order of 0.5 kcal per kilogram per kilometre above resting — for a 70 kg person, a 5 km walk is somewhere near 175 kcal net, not the 300 you might have hoped.
None of this makes the devices useless. They are decent at telling you whether today was busier than yesterday. They are not scales for food.
Flat target or daily adjustment?
A flat target absorbs your training into one number you eat every day, gym or no gym. It is simpler, it removes the daily negotiation with your watch, and it stops the pattern where a big session licenses a big evening. It suits most people, particularly those whose week looks broadly like the last one.
Daily adjustment raises intake on hard days and lowers it on rest days. It makes sense when the variation between days is genuinely large — a Sunday long run of 25 km, a rest week after a race, shift work where some days are on your feet for nine hours and others are not. Trying to average a 2,000 kcal training week against a 600 kcal one into a single flat figure leaves you overfed half the time and underfed the other half.
If you do adjust, most people who do it successfully add back a fraction — a half is the common rule of thumb — and treat it as a hedge against an optimistic number, not as a calculation. There is nothing scientific about halving it. It is just an admission that the figure is soft.
The version of this that avoids the problem entirely
All of the above is arithmetic about estimates. There is a way round it: measure the number instead.
If you take what you actually logged over a fortnight and compare it against what the scale actually did across the same fortnight, the result already contains your training, your fidgeting, your commute and your slow or fast metabolism. It does not need to know what a spin class costs, because the spin class is inside the weight data. Day So Far builds its target this way rather than from an activity multiplier, which is why it does not ask you to add sessions on top.
The catch is honest: the method depends on your logging being roughly consistent. If you log carefully in the week and vaguely at weekends, a measured maintenance figure will come out too low, and no clever maths fixes it.
Is there a real case for eating more around hard training?
Yes, and it has little to do with balancing a ledger. Long or intense sessions on very little food tend to feel bad, go badly, and are often followed by eating far more than you saved. Appetite after a hard effort is information, not a failure of will. Eating somewhat more on those days — and in particular getting some carbohydrate and protein around the session — is a reasonable thing to do, and most people arrive at it without any calculation at all.
That is not a training-nutrition plan, and if you are preparing for something specific, someone qualified should write you one.
And do not let this put you off training
Exercise is a poor calorie eraser and an excellent almost everything else. It is how you keep muscle while losing fat, and it does things for sleep, mood, blood pressure and how strong you feel at seventy that no deficit does. Judging a session by the number on your wrist is judging it by the one thing it is worst at.
If the scale is not moving, the productive move is not to argue with the watch. It is to hold intake steady for two or three weeks, weigh often enough to see through the daily noise, and change the target based on what actually happened.