What should I eat to avoid the 3pm energy crash?

Half past two. You have read the same paragraph three times. Your eyes are hot, the meeting at three feels like a physical obstacle, and you are thinking about the vending machine with a clarity you did not bring to any of this morning's work.

Almost everyone who works at a desk knows this window. What most people do not know is that roughly half of it has nothing to do with what they ate, and treating the whole thing as a food problem is why the fixes keep failing.

This chapter separates the two halves, and gives you something to do about each.

The uncomfortable part: it is not entirely your lunch#

The afternoon dip in alertness has been studied properly, and the results are not what the phrase "post-lunch dip" implies.

Researchers have shown that the dip appears even in people who have eaten no lunch at all, and in people who do not know what time it is. It tracks a roughly twelve-hour harmonic in the circadian system, a second smaller trough sitting opposite the main one that puts you to sleep at night. Reviews of the research describe it as largely unrelated to lunch, and note that a poor night's sleep makes it considerably worse.

Why it works. Your alertness is not a fuel gauge. It is a rhythm, and the early afternoon is a scheduled low point in it. Food can deepen the trough or leave it alone, but food did not create it.

How you apply it at a desk. Stop trying to eat your way out of the whole thing, and start scheduling around it. Put the work that needs judgement in the morning and the work that needs hands in the afternoon. That single change does more than any lunch.

The half that is your lunch, with numbers#

Sarah Berry, a nutrition researcher at King's College London, co-leads PREDICT, a large programme studying how individuals respond to identical meals in real life. One of its findings, published in Nature Metabolism in 2021, is directly about this window.

The team tracked continuous glucose in around a thousand people across the UK and the United States, through more than 8,000 standardised meals and 70,000 ordinary ones. What predicted hunger afterwards was not the peak in blood sugar. It was the dip two to three hours later, when glucose fell below where it started.

People with large dips, whom the researchers called big dippers, reported around 9 percent more hunger, waited about half an hour less before their next meal, and ate roughly 312 more calories over the course of the day than people with small dips. They had eaten exactly the same meals.

Why it works. A meal that arrives fast gets a large insulin response, which can overshoot and pull glucose below baseline. That undershoot is what you feel at 3pm as fog and hunger, and it is a rebound rather than a shortage.

How you apply it at a desk. You cannot flatten the curve by eating less. You flatten it by changing what arrives: protein, fat and fibre alongside the carbohydrate, and wholegrain rather than refined. The lunch that avoids the dip is not a smaller lunch, it is a slower one.

Why the enormous lunch backfires twice#

Portion size matters here for a second, mechanical reason, separate from the blood sugar story.

A very large meal takes a long time to digest, and digestion itself is associated with drowsiness. So a big refined-carbohydrate lunch does both things at once: it produces the glucose swing the PREDICT research describes, and it adds a heavy digestive load on top of a circadian trough that was arriving anyway. Three effects, one meal, all pointing the same direction.

Why it works. The three causes stack rather than compete, which is why the giant Friday lunch feels categorically worse than a slightly larger normal one rather than proportionally worse.

How you apply it at a desk. Aim for satisfied rather than full at lunch, and put the volume into vegetables rather than into the carbohydrate. If the afternoon matters, the enormous lunch is a choice with a known cost.

The coffee that fixes today and breaks tomorrow#

The standard response to the 3pm crash is caffeine, and it works. It also has a cost that most people have never seen quantified.

Christopher Drake and colleagues at the Henry Ford Hospital sleep centre gave healthy sleepers 400 milligrams of caffeine, roughly two to three cups of coffee, at bedtime, three hours before bed, and six hours before bed. All three disrupted sleep. The six-hour dose, which is a 4pm coffee for someone going to bed at 10pm, cut objectively measured total sleep time by more than an hour.

The detail that matters most: the participants did not notice. Their own sleep diaries did not register the disturbance the monitors recorded.

Why it works. Caffeine blocks the receptors that register adenosine, the molecule that builds up through the day and creates sleep pressure. The pressure keeps accumulating; you simply stop feeling it. When the caffeine clears, it is all still there, and its half-life of around five hours means a late dose is still working at bedtime.

How you apply it at a desk. Draw a caffeine line about eight hours before bed, so early afternoon for most people. A poor night makes tomorrow's dip worse, which is how a 4pm coffee becomes a self-renewing subscription.

The anti-crash protocol#

Two parts: the lunch, and the twenty minutes around three o'clock.

1. 1Build lunch on protein, roughly 30 grams. The target from Chapter 1 of this series. Chicken, fish, eggs, prawns, tinned fish, or the pulse-based version from the Veg Food series. This is the component that slows everything else down.
2. 2Make the carbohydrate wholegrain, and moderate. Brown rice, wholegrain bread, whole wheat pasta. A quarter of the plate rather than the base of it, which is the plate rule from the Healthy Food series applied to the specific problem of an afternoon.
3. 3Add fat and volume. Olive oil, nuts, avocado, and half a plate of vegetables. Fat slows gastric emptying, fibre slows absorption, and volume fills you at a size that does not make you sleepy.
4. 4Stop at satisfied. The gap between comfortably fed and heavily fed is the difference between a workable afternoon and a lost one.
5. 5Set the caffeine cutoff. Roughly eight hours before bed. If you need the 3pm coffee to function, note that it is buying today's afternoon with tomorrow's.
6. 6Then move for ten minutes at the dip. A walk outside beats a walk inside, because daylight is the strongest signal your circadian system takes. This is the half of the problem that food cannot solve, and it is the half most people never address.

The walk is not a consolation prize. Given that a meaningful share of the dip is circadian, light and movement are aimed at the actual mechanism, and the lunch is aimed at the rest.

The individual variation nobody tells you about#

One more finding from the PREDICT work, because it changes what you should expect from any advice on this subject, including this chapter.

The participants ate identical standardised meals, and their blood sugar responses were not identical. Some people dipped substantially after a given meal and others barely moved. That variation between people was large, and it means the food that flattens your afternoon is not necessarily the food that flattens someone else's.

Why it works. Response to a meal depends on more than the meal: sleep the night before, what you ate at the previous meal, the time of day, and whether you moved afterwards all shift it. A population average is a starting point, not a prescription.

How you apply it at a desk. Run it as a two-week experiment rather than taking it on faith. Same lunch structure for a fortnight, and note the afternoon on a scale of one to five. If nothing moves, the food half was not your bottleneck and the sleep or the daylight half probably is.

What the ten-minute walk is doing#

The walk earns its place for three separate reasons, and it is worth knowing them because it is the step most likely to feel like filler.

Daylight. Outdoor light, even on an overcast day, is many times brighter than office lighting and is the main signal your circadian system reads. It is the only lever in this chapter that acts on the circadian half of the problem directly.

Muscle contraction. Working legs pull glucose out of the bloodstream with little insulin involved, which is the mechanism behind the walking-break research in Fitness Chapter 1. A short walk after lunch blunts the swing rather than waiting for it.

A break from the screen. Some of what feels like an energy crash at 2:40pm is four hours of unbroken attention. Ten minutes away from a screen is a different intervention from ten minutes of daylight, and you get both at once.

What trips people up#

Skipping lunch entirely. A popular fix that trades a 3pm dip for a 5pm collapse and an enormous dinner. The circadian trough arrives whether or not you have eaten, so the fasted version is the same dip with less to work with.

The mid-afternoon sweet fix. A biscuit at three is a fast carbohydrate on top of an existing dip, which is the exact pattern the PREDICT research associates with more hunger and more calories later. If you need something, make it protein: yoghurt, nuts, a boiled egg.

Blaming the lunch for all of it. People try five different lunches, still feel the dip, and conclude that nothing works. Some of it is not food and never was. Expect improvement, not elimination.

Ignoring last night. A short or broken night deepens the afternoon dip measurably. If you are crashing hard every day, the intervention with the highest return is probably an earlier bedtime rather than a better sandwich.

Persistent, severe afternoon fatigue that does not respond to better sleep and food is worth investigating. Anaemia, thyroid problems, sleep apnoea, diabetes and depression all present this way, and all are treatable. See a doctor rather than reorganising your lunch for another six months.