Sugar cravings before your period: the measured figure
is 168 kilocalories, not a few hundred

Three different things,
measured with three instruments

Put side by side the first pages that answer this question in Romanian, and they all say the same sentence: hormones change, metabolism goes up, so you eat more and crave sweet things. The sentence is not false. The problem is that it ties into a single chain three things that are measured separately, with three different instruments and three very different degrees of certainty.

The first is the craving. It is a sensation, and the only way to find out about it is to ask. It is recorded in diaries, on scales from zero to ten, and the result depends on who asks, how they ask and what someone remembers about last week.

The second is intake, meaning the energy that goes in. It is counted from food diaries, from weighing or, in the best studies, from meals served in the lab. In most cases it is still self-reported, but at least it has a unit of measurement.

The third is resting expenditure, meaning the energy you burn while sitting still. Here you no longer ask anything: you put on a mask or a hood and measure oxygen. It is the most objective of the three.

The surprising part is that the certainty of the figures does not follow this order. The most objective of the three is precisely the one that shows the smallest difference, and the one least written about. It is the same trap as with bromelain and papain, where the figure everyone reads is not the figure that tells you anything.

How much more you actually eat,
in the second half of the cycle

In July 2024 the first systematic review with meta-analysis on this question appeared. The authors gathered fifteen datasets, 330 participants, a mean age of 26 with a standard deviation of 4, and a mean body mass index of 22.4. The inclusion criteria required every study to have measured both intake and cycle phase.

The result: intake was higher in the luteal phase, with a standardised effect size of 0.69 and a p of 0.039. And the raw difference, expressed in what you put on your plate, was 168 kilocalories a day.

This figure does not appear in any of the Romanian articles on the first page. Instead you find “a few hundred extra calories”, with no source at all. The difference between the two phrasings is not a matter of style. 168 is a snack. “A few hundred” is a meal, and between the two sits the feeling that you have lost control.

Two things need to be said in the same breath. First: heterogeneity across the included studies was high, an I squared of 83%, which means the studies did not find the same thing. Second: the authors themselves close their review by saying they came across “repeated methodological inconsistencies” and that future work should use better practice, both in measuring intake and in establishing cycle phase.

A narrative review published a year earlier, in the same journal, had put the range on paper: the mean increases reported by studies run from 90 to 504 kilocalories a day, with one study at 529 and another at 337. The authors' conclusion is the most honest sentence in this whole literature: the size of the fluctuations “is not yet clearly quantifiable” and most likely varies both between women and from one cycle to the next in the same woman.

Do the division yourself. 168 over 504 gives a third. The figure you read everywhere is the top end of a range in which you could sit anywhere.

The metabolism part is the smallest,
and the shakiest

For the metabolism part there is a separate meta-analysis, of 26 studies and 318 women in total, published in 2020. It measured resting metabolic rate in the follicular and the luteal phase.

The result, in the authors' words: a small but significant effect in favour of a higher rate in the luteal phase, with an effect size of 0.33 and a confidence interval of 0.17 to 0.49. Heterogeneity was very low, an I squared of 3.8%, and the test for publication bias came out clean. The authors received no funding for the work.

Then they did something worth seeing. They reran the analysis only on studies published from 2000 onwards, that is, those with modern equipment and better verified phases. There the effect size dropped to 0.23, and the confidence interval stretched from zero to 0.47, with a p of 0.055.

An interval that touches zero means, in plain language, that the difference can no longer be told apart from its absence. Not that it does not exist. That it can no longer be seen, with good methods, in the number of women studied so far.

That is all that stands behind the sentence “metabolism rises by 2 to 10%” that you find on several Romanian sites, with no source and no interval. The objectively measured part of the whole story is also the smallest and the least certain.

The low point is not where you think,
the dip is at ovulation

Here is the most curious gap in everything written on the subject. Every article talks about a rise in the luteal phase, as though the rest of the cycle were a straight line. It is not.

The 2023 narrative review says it plainly: intake appears lower in the follicular phase than in the luteal phase, with a marked drop in the days leading up to ovulation and on the day of ovulation. In other words, the lowest point of the cycle is not at its start. It is in the middle.

The practical consequence is immediate. If you compare the week before your period with the week around ovulation, you are comparing two extremes and get the largest possible difference. If you compare it with the first week after your period, you get a smaller difference. Same woman, same cycle, two different figures, just because you picked a different starting point.

The hormonal hypothesis behind this is written in the literature as a hypothesis, not as fact: oestrogen would suppress appetite, and progesterone, in the presence of oestrogen, would do the opposite. Around ovulation oestrogen is high and progesterone almost absent. In the luteal phase the ratio reverses. The fit is neat, but the wording in the reviews stays in the conditional, and it is right that it does.

How long does this window last? Across 327 luteal phases from ovulatory cycles, measured from the luteinising hormone peak to the day before menstruation, the mean length was 14.13 days, with a standard deviation of 1.41 days. Short luteal phases, under ten days, occurred in 5.2% of cycles.

So the window is almost two weeks, not five or seven days. And if your craving comes later or earlier than your friend's, the two days of standard deviation are already explanation enough, with nothing wrong with either of you.

Chocolate is a separate story,
and it is largely learned

So far we have talked about energy in general. The craving for chocolate is something else, and it has the most elegant study in the whole file.

In 2017, two researchers surveyed 275 women about their cravings and their background. They split the sample into three: women born in the United States to American parents, 101 people; daughters of immigrants, born there, 93 people; and women born in another country, 81 people.

Across the whole sample, a third said they had a chocolate craving linked to their cycle, 85 women out of 275, or 30.9%. By group, the figures split apart: 17.3% among women born outside the United States, 32.7% among American women and 40.9% among daughters of immigrants. The difference between the first group and the other two was statistically significant.

Stop on this for a second. The hormones are the same in all three groups. The luteal phase lasts just as long. If the chocolate craving before your period were a strictly hormonal phenomenon, the three percentages should be equal. They are more than two to one apart at the extremes, and the group that reports it most often is the one raised in the culture where this story is told the most.

For context, the same authors note that chocolate cravings in general are reported by over 90% of women in the United States, and for about half of them they intensify around menstruation. So the phenomenon is real and very widespread. It is just that its link with the cycle seems to be learned rather than inherited.

What do you do with this information? You do not use it to second-guess yourself. You use it to separate two questions that are always mixed together: “does my body want more energy?”, to which the measured answer is yes, by 168 kilocalories, and “does my body want chocolate specifically?”, to which the answer is far less clear.

Serotonin: what was actually tested,
and on how many people

The serotonin explanation appears in every article on this topic, usually in two sentences and with no reference at all. It is worth seeing in its original form, because it was actually tested.

The starting idea is a real piece of biochemistry: a meal rich in carbohydrate and low in protein raises, in the blood, the ratio of tryptophan to the other large neutral amino acids. Tryptophan is the raw material serotonin is made from, and all these amino acids enter the brain through the same transporter, so the ratio matters, not the absolute amount.

In 1995, a team built a drink formulated specifically to do exactly that and tested it on 24 women with confirmed premenstrual syndrome. A double-blind, crossover study spread over three cycles, preceded by a month-long placebo run-in. Measurements were taken at home or at the office, through an interactive computer and telephone system, before the drink and at 30, 90 and 180 minutes after, in the late luteal phase.

The result: 90 to 180 minutes after the drink, self-reported scores for sadness, irritability and confusion fell significantly, and so did carbohydrate craving. The most important part is the comparison: the other two products, with exactly the same calorie intake, did not significantly change anything.

That is good information, because it shows it was not just a matter of getting calories into someone. And it is, at the same time, small information: 24 women, a drink formulated in a lab, a three-hour window, a single study. It is not a recipe, it is not a recommendation and it is not a reason to buy anything.

What is left of it, honestly: the mechanism exists and was measured once, on very few people. If you feel better after something sweet in the second half of your cycle, you are not imagining it. But there is no figure that says how much, from what, and for whom. It is the gap between a mechanism verified in the lab and a measurement in someone's blood, the same gap we wrote about with quercetin and vitamin C.

What to do with this in practice,
without fighting yourself

Nothing that follows is about fighting the craving. Four things, all about measurement.

Compare a day with the same day, not a week with a week. If you compare the day before your period with a day around ovulation, you exaggerate the difference, because that is where the cycle's minimum sits. The honest point of comparison is the same day of the cycle, in two different months.

Put the snack in your schedule, at the time the craving usually arrives. 168 kilocalories is a fruit yoghurt, two squares of dark chocolate with a handful of nuts, or a slice of bread with honey. If you know it catches you around nine in the evening, nine o'clock is information, not an enemy. A snack put there in advance is not giving in, it is a plan.

Write down two cycles, not one. The literature says the size of the change varies from one cycle to the next in the same woman. A single recorded cycle does not give you a pattern, it gives you an anecdote. Two or three show you whether you actually have a rhythm.

Do not mistake the range for a target. Almost anything fits between 90 and 504 kilocalories a day. Your figure is yours, and other people's average is not a norm you have to reach or stay under.

One thing to say directly, so it does not hang in the air. There is no supplement that is the answer to this. Nobody has one, and it is not what we sell. What can be useful around the subject is something else entirely: what happens with bloating before your period, where the measured peak is on the first day of flow and not before, and how to tell a bloated belly from belly fat with a tape measure, which is the same idea applied to a different measurement. And if you have come across the magnesium deficiency explanation, which circulates widely on the subject of cravings, the arithmetic in the article on magnesium bisglycinate shows why the figure on a magnesium label is not the figure for magnesium.

Frequently
asked questions