The same food, a worse response — and melatonin is the reason#
Eat an identical meal late and your blood glucose climbs higher and clears more slowly than it would have earlier. The clearest measurement of this gave 845 adults two identical 75 g glucose loads on separate evenings, each after an 8-hour fast: one scheduled four hours before their habitual bedtime (averaging 19:49), the other one hour before it (averaging 22:48). In the late condition, circulating melatonin was 3.5-fold higher — 66.18 pg/mL against 22.76 — and the same glucose load produced 8.3% higher glucose area under the curve on 6.7% less insulin1.
That is the answer, and the mechanism inside it is more useful than the headline. The exposure that matters is not the hour on the clock — it is whether your food overlaps with the melatonin your body releases as it prepares for sleep. Melatonin suppresses the insulin your pancreas puts out in response to a meal, so a carbohydrate load arriving after the melatonin has risen is met with a weaker insulin response than the same load arriving before it. And how much that costs you depends partly on a common variant in the gene for the melatonin receptor on your pancreatic beta cells.
Note what this article is not claiming. Nothing here says a late meal contains more energy or is stored differently; that separate question is settled in does eating late at night make you gain weight, and the clock's own contribution — the endogenous circadian effect, separated from meal timing and from misalignment — belongs to your body clock and how it times metabolism. This page is about the glucose response, and specifically about the hormone driving it.
Why anchoring to bedtime is the right way to ask the question#
One design detail carries a lot of weight. Garaulet's two conditions were not "7pm versus 11pm." They were set relative to each participant's own habitual bedtime, and melatonin was measured directly in both. That matters because 10pm arrives at very different points in different people's biological night, so a study built on a fixed clock hour averages together people whose melatonin has been rising for an hour and people whose melatonin hasn't started — a problem the late-eating literature runs into repeatedly.
By anchoring to bedtime and then confirming a 3.5-fold melatonin difference, this trial measured the exposure the mechanism actually cares about rather than a proxy for it. That is why its effect is visible in 845 free-living Spanish adults rather than needing a sealed laboratory.
Give melatonin in the morning and the same thing happens#
An association between late eating and worse glucose could always be something else about evenings — you have been awake longer, you have eaten already, you are less active. The way to isolate melatonin is to administer it when there isn't any.
That study gave 21 healthy women 5 mg of melatonin or placebo fifteen minutes before a 75 g glucose tolerance test, on four non-consecutive days, at 9am and at 9pm. Melatonin impaired glucose tolerance at both times. In the morning it raised the incremental glucose area under the curve by 186% and peak glucose by 21%; in the evening, by 54% and 27%. The mechanism differed by time of day — the morning impairment came from reduced insulin secretion, the evening one from reduced insulin sensitivity2.
Melatonin degraded the glucose response at 9 in the morning, when nobody's body is producing any. The problem was never the hour. It was the hormone meeting the carbohydrate.
The morning effect being the larger one looks backwards until you consider what changed in each case. At 9am endogenous melatonin is essentially absent, so 5 mg is an enormous relative increase; at 9pm the participants' own melatonin was already rising, so the same dose is a smaller proportional change against a higher baseline. That reading is our inference from the design rather than a claim the paper makes — but it fits, and it means the evening figure is the conservative one.
| The exposure | What happened to glucose | Mechanism identified |
|---|---|---|
| 5 mg melatonin + 75 g glucose, 9am2 | Incremental AUC +186%, peak +21% | Reduced insulin secretion |
| 5 mg melatonin + 75 g glucose, 9pm2 | Incremental AUC +54%, peak +27% | Reduced insulin sensitivity |
| 75 g glucose 1 h vs 4 h before bed1 | AUC +8.3% on 6.7% less insulin | Reduced beta-cell function |
| Nightly bedtime melatonin, taken without food4 | Fasting glucose essentially unchanged | — |
Whether it costs you much depends on a gene#
The melatonin receptor on your pancreatic beta cells is built from a gene called MTNR1B, and a common variant of it changes how much the receptor's signal matters. In Garaulet's trial the late-versus-early penalty was significantly larger in carriers of the G risk allele: each copy added 12.53 mg×h/dL to the glucose-AUC difference (P for interaction = 0.009) and subtracted 9.56 µU×h/mL from the insulin-AUC difference (P = 0.035), with the disposition index — a summary of beta-cell capacity — moving in step (P = 0.018)1.
Why that variant specifically is not a mystery. Across 19,605 Europeans, the same G allele raised the odds of isolated impaired fasting glycemia (OR 1.64) and was associated with decreased insulin release after both oral and intravenous glucose challenges3. These are people whose meal-time insulin response is already the weaker link. Late eating leans on precisely the mechanism they have least of.
The variant is common enough that an 845-person trial could split cleanly on it, which means a meaningful minority of readers carry at least one copy. Two things follow, and they pull in opposite directions. This is a real gene-by-timing interaction — a case where identical advice genuinely has different value for different people. It is also not something you should go and get tested for: the outcome is a glucose curve, not a diagnosis, and the behavioral response is the same one that suits everybody, just with more to gain.
The melatonin supplement question, which is a different exposure#
If swallowing melatonin fifteen minutes before a glucose load wrecks the response, is a nightly melatonin tablet damaging your metabolism? Pooled across randomized trials of at least two weeks of daily bedtime melatonin in healthy adults and people with metabolic disease, fasting glucose was essentially unchanged (effect size −0.07), fasting insulin fell slightly (−0.27), insulin resistance trended down (−0.20), and HbA1c was unaffected. The authors concluded melatonin looks like a glucose-metabolically safe sleep aid, while flagging wide confidence intervals4.
Those two results are not in conflict, and what separates them is the whole point of this article: co-incidence with food. A bedtime tablet taken hours after dinner delivers melatonin into a fasted body, where there is no glucose load for the weakened insulin response to fail against. The acute experiment deliberately delivered melatonin and 75 g of glucose together. Rubio-Sastre's authors drew exactly that conclusion themselves — that when administering melatonin, proximity to meal timing may need to be considered, particularly for those at risk of glucose intolerance.
So the practical translation is narrow and specific: if you take melatonin, don't take it alongside a meal or a snack.
What this changes, and what it doesn't#
Start with what it doesn't. A bigger glucose curve is not more calories. The energy in a late dinner is identical to the energy in the same dinner four hours earlier, and none of the numbers above describe fat gain — they describe how quickly your body clears a sugar load. Whether meal timing reaches the scale at all is a separate and much more modest question, handled in does meal timing affect fat loss.
What it does change is the shape of a sensible rule. "Don't eat after 8pm" is keyed to the wrong variable, because 8pm sits at different points in different people's melatonin curve. The mechanism supports something you can actually personalize: leave a gap between your last carbohydrate-heavy meal and the time you go to bed. Roughly three to four hours is the separation that distinguished the two conditions in the trial, and it works for a night owl and a morning person alike because it is measured against your bedtime rather than the wall.
Two caveats before you act on it. Both key studies used a 75 g glucose drink, which is a harsher and faster carbohydrate load than most dinners — a mixed meal with fat, protein and fiber empties more slowly, which is part of why what the carbohydrate arrives with matters. And this all sits alongside a larger effect from the same cluster: losing sleep hits insulin sensitivity harder than shifting your dinner does, which is walked through in how one bad night hurts insulin sensitivity. If both a late dinner and a short night are on offer, the sleep is the bigger lever — and the hormonal cascade tying the two together is audited in the cluster's pillar.
FAQ#
Does taking melatonin with food affect your blood sugar?#
Acutely, yes, and the effect is not small. Five milligrams of melatonin taken fifteen minutes before a 75 g glucose load raised incremental glucose area under the curve by 186% in the morning and 54% in the evening2. Taken at bedtime on an empty stomach, though, daily melatonin left fasting glucose essentially unchanged across pooled randomized trials4. The separation matters more than the supplement: keep melatonin away from meals and snacks.
Is a late dinner worse for some people than others?#
Measurably so. In an 845-person crossover trial, carriers of the MTNR1B G risk allele took a significantly larger hit from eating one hour before bed instead of four — an extra 12.53 mg×h/dL of glucose AUC per copy of the allele, traced to reduced beta-cell function1. The same variant is independently linked to impaired fasting glucose and weaker insulin release3. Genetic testing isn't the useful response; leaving a gap before bed is, and it costs nothing if you turn out to be a non-carrier.
Does a bigger glucose spike at night mean the meal had more calories?#
No. The two experiments here used an identical glucose load in both conditions — the energy was fixed by design, and only the timing moved. What changed was how the body handled it: less insulin released, glucose clearing more slowly1. That is a metabolic-handling difference relevant to long-run glucose control, not an arithmetic one relevant to your daily total.
Sources#
- Garaulet M, Lopez-Minguez J, Dashti HS, et al. Interplay of dinner timing and MTNR1B type 2 diabetes risk variant on glucose tolerance and insulin secretion: a randomized crossover trial. Diabetes Care. 2022;45(3):512-519.
- Rubio-Sastre P, Scheer FAJL, Gómez-Abellán P, Madrid JA, Garaulet M. Acute melatonin administration in humans impairs glucose tolerance in both the morning and evening. Sleep. 2014;37(10):1715-1719.
- Sparsø T, Bonnefond A, Andersson E, et al. G-allele of intronic rs10830963 in MTNR1B confers increased risk of impaired fasting glycemia and type 2 diabetes through an impaired glucose-stimulated insulin release: studies involving 19,605 Europeans. Diabetes. 2009;58(6):1450-1456.
- Lauritzen ES, Kampmann U, Smedegaard SB, Støy J. Effects of daily administration of melatonin before bedtime on fasting insulin, glucose and insulin sensitivity in healthy adults and patients with metabolic diseases: a systematic review and meta-analysis. Clin Endocrinol (Oxf). 2021;95(5):691-701.
