The NHS limit for caffeine in pregnancy is 200 mg a day. A newborn takes around four days to clear a single dose, the largest meta-analysis on the subject found no level below which the risks disappear, and one high street espresso has been measured at 322 mg.
Those three facts sit uneasily next to each other. This post walks through where the 200 mg figure came from, what caffeine does inside a pregnancy, what the studies found, and what the people arguing the limit is fine have on their side. It ends with what we would do.
Contents
- Where the 200 mg figure came from
- What pregnancy does to clearance
- The baby cannot clear it at all
- What the studies found
- The case for leaving the limit alone
- Why 200 mg is harder to stay under than it looks
- What to do with this
The short version
Caffeine crosses the placenta freely and the fetus has none of the liver enzyme that breaks it down. A pregnant woman's half-life stretches from about 6 hours to as much as 15 in the third trimester, and a newborn's runs to 80 to 100 hours. Meta-analyses of more than 50 studies find risk rising with every extra 100 mg a day and no threshold underneath it. The 200 mg limit is where older studies stopped being able to see harm, which is a different thing from where harm stops.
Where the 200 mg figure came from
Until 2008 the UK advice was 300 mg a day. The Food Standards Agency cut it to 200 mg that year on the advice of its toxicity committee, after a government-funded study found fetal growth restriction climbing with intake and no obvious floor (BMJ, 2008). The European Food Safety Authority settled on the same figure in 2015. The American College of Obstetricians and Gynecologists uses it too.
The figure is not a measured safe dose. Nobody has run a trial that found harm at 250 mg and none at 200. The number marks the point below which the studies available at the time could no longer show a statistically clear effect, in cohorts that were mostly too small to see a modest one.
A 2020 review in BMJ Evidence-Based Medicine went through 48 observational studies and meta-analyses from the previous two decades. Of 42 sets of findings, 32 showed increased risk. Of 17 meta-analyses, 14 agreed on miscarriage, stillbirth, low birth weight and childhood leukaemia. The author's conclusion was that current advice is not supported by the evidence and that pregnant women should be told to avoid caffeine (BMJ Evidence-Based Medicine, 2020).
200 mg is a ceiling that was set by what the data could prove in 2008. It was never a floor under which the baby is known to be fine.
What pregnancy does to clearance
Caffeine leaves the body through one liver enzyme, CYP1A2, and pregnancy hormones slow it down. In a normal adult the half-life is about 6 hours, which is the figure our caffeine calculator uses. In pregnancy it is not.
- A saliva study of 57 pregnant women found a mean half-life of 8.3 hours with a range of 3 to 16, against 3.4 hours in the non-pregnant adults measured alongside them. Values returned to normal within a month of delivery (European Journal of Clinical Pharmacology, 1981).
- In the last two weeks of pregnancy, clearance was 39% of the non-pregnant rate. In the same women it more than tripled again within twelve weeks of the birth (Canadian Medical Association Journal, 1982).
- A 2025 review puts the third-trimester half-life at around 15 hours (Nutrients, 2025).
Run the numbers on a single 200 mg coffee, drunk at 8am and never topped up.
| Time since the coffee | Adult, 6 h half-life | Third trimester, 15 h | Newborn, 96 h |
|---|---|---|---|
| Bedtime, 15 hours | 35 mg | 100 mg | 180 mg |
| Next morning, 24 hours | 12.5 mg | 66 mg | 168 mg |
| Three days, 72 hours | under 1 mg | 7 mg | 119 mg |
Read the middle column. A woman in her third trimester who has one coffee at breakfast goes to bed with 100 mg still circulating, and wakes up with 66 mg of yesterday's coffee on board before today's. Every drink stacks on the last, the same arithmetic we mapped in how long caffeine stays in your system, except that in late pregnancy the curves are two and a half times longer.
At a 15-hour half-life, a daily 200 mg coffee settles at a steady level about 50% higher than the coffee itself. The 200 mg limit is measured against what goes in. The fetus is exposed to what stays.
The baby cannot clear it at all
The mother at least has a slowed-down enzyme. The fetus has none.
- Caffeine crosses the placenta freely and fetal serum concentrations run close to the mother's (European Journal of Epidemiology, 2014).
- CYP1A2 is not detectable in the liver for the first 30 days of life. In the first month, more than 85% of caffeine leaves a baby's body unchanged in urine, because there is little or no metabolism to speak of (Clinical Pharmacology and Therapeutics, 1979).
- The resulting half-life in a newborn is roughly 100 hours. Reviews put it at 80 to 100 hours in term babies, and it can run longer in premature ones (Seminars in Fetal and Neonatal Medicine, 2020).
- Adult clearance is not reached until three to four and a half months of age (Archives of Disease in Childhood, 1979).
Look at the right-hand column in the table again. Three days after a single coffee, a newborn still has 119 mg of the 200 in its system. Neonatal units use caffeine as a drug to treat breathing pauses in premature babies, and they dose it once a day precisely because it hangs around so long.
The same figure governs breastfeeding. Caffeine passes into milk, and a baby who cannot clear it accumulates it. With a four-day half-life, about a third of Monday's dose is still present the following Sunday, underneath everything drunk since.
A dose that is gone from you in a day and a half is still mostly there in your baby after three. That is the fact the 200 mg limit does not mention.
What the studies found
None of this would matter if the exposure were harmless. The evidence says otherwise, and it says so with a dose-response, which is the pattern that makes epidemiologists sit up.
Per 100 mg a day
The largest dose-response meta-analysis pooled 53 cohort and case-control studies. For every extra 100 mg a day, roughly one cup, it found risk of miscarriage up 14%, stillbirth up 19%, low birth weight up 7% and small-for-gestational-age up 10%. Preterm birth did not move. Its exact words on a threshold: there is no identifiable level below which the associations are not apparent (European Journal of Epidemiology, 2014).
A separate meta-analysis of 13 prospective studies covering over 90,000 women put the low birth weight figure higher, at 13% per 100 mg, and found the risk already raised at 50 to 149 mg a day (BMC Medicine, 2014).
Under the limit
- The UK CARE study behind the 2008 change: fetal growth restriction odds 1.2 at 100 to 199 mg, 1.5 at 200 to 299 mg, 1.4 above 300 mg, with a clear trend and no floor. It also found the effect was stronger in women who cleared caffeine faster, which points at the metabolites and not only the caffeine (BMJ, 2008).
- A Kaiser Permanente cohort found miscarriage risk more than doubled at 200 mg or more a day, hazard ratio 2.23, and a raised but statistically uncertain 1.42 below it. Controlling for morning sickness made no difference (American Journal of Obstetrics and Gynecology, 2008).
- The US NICHD Fetal Growth Studies measured caffeine in the blood of 2,055 non-smoking women at 10 to 13 weeks. The top quarter delivered babies 84 g lighter, shorter, and with smaller head, arm and thigh circumference. Women reporting about 50 mg a day, half a cup, already showed the difference (JAMA Network Open, 2021).
Years later
- The same cohort followed the children to age 4 to 8. Those whose mothers were in the top quarter for blood caffeine were shorter, and in a 1960s cohort with heavier intake the height gap widened every year to age 8 (JAMA Network Open, 2022).
- Brain scans of 9,157 nine and ten year olds in the ABCD study found altered white matter in two major fibre tracts with prenatal caffeine exposure, and worse scores on every behavioural measure tested, though no measurable effect on cognition. The authors' line: current guidelines may require recalibration (Neuropharmacology, 2021).
A blood-measured cohort of 2,000 women and a meta-analysis of 90,000 both saw differences under 150 mg a day. The limit is 200.
The case for leaving the limit alone
The people who set the limit are not fools, and they have real objections.
- Almost none of this is randomised. The Cochrane review found exactly one usable trial. It gave 1,197 Danish women caffeinated or decaffeinated instant coffee from mid-pregnancy, cut intake by an average 182 mg a day, and saw no difference in birth weight or prematurity (Cochrane Database of Systematic Reviews, 2015). One trial that started after the first trimester is thin evidence, but it is the only experiment there is and it came back null.
- Healthy pregnancies make women nauseous, and nauseous women drink less coffee, so the pregnancies that were going to fail anyway are over-represented among the coffee drinkers. This is the strongest single objection. The Kaiser study adjusted for it and the association held.
- Coffee drinkers smoke more, drink more alcohol and eat differently. Adjustment helps and never fully fixes it.
- The 2014 meta-analysis flagged small-study effects, a sign of publication bias, in its own data. It concluded that the associations were modest within usual intake and potentially explained by bias, and recommended keeping the 200 mg limit where it is.
Both readings of the same data are defensible. Where they split is on what to do while the uncertainty lasts.
If the sceptics are right, a woman who cuts caffeine has given up coffee for nine months for nothing. If they are wrong, a woman who kept to the limit has spent nine months at a dose the studies associate with a lighter, shorter baby and a higher chance of losing the pregnancy. Those are not symmetrical outcomes, and a limit set at the edge of what the data could prove leans the wrong way.
The cost of being too cautious is a decaf. The cost of not being cautious enough does not have a price.
Why 200 mg is harder to stay under than it looks
The limit also assumes you know what is in your cup, and most people do not.
The NHS list gives 100 mg for a cup of instant coffee, 75 mg for tea, 80 mg for a 250 ml energy drink and 40 mg for a can of cola. Those figures are fine for a kitchen. They fall apart in a coffee shop.
Researchers at Glasgow bought 20 single espressos from high street coffee shops and measured them. The caffeine content varied six-fold. The strongest was 322 mg, and three others were over 200 mg, meaning four of the twenty single shots broke the pregnancy limit on their own (Food and Function, 2012). A flat white or latte is normally built on a double.
| Drink | Caffeine | Share of 200 mg |
|---|---|---|
| Cup of tea (NHS figure) | 75 mg | 38% |
| Cup of instant coffee (NHS figure) | 100 mg | 50% |
| High street single espresso, measured range | 51 to 322 mg | 25% to 161% |
Add the two teas most people do not count, the square of dark chocolate and the cola with lunch, and a woman who thinks she is having "one coffee a day" can be well past 300 mg without a single decision that felt like a risk. We covered the same blind spot in energy drinks vs coffee. Tea also contains the amino acid L-theanine, and what the L-theanine research shows is a separate question from how much caffeine is in the cup.
What to do with this
We sell a caffeine taper, so treat what follows as a lay reading of the evidence and talk to your midwife or GP before acting on any of it.
If you are pregnant
- Treat 200 mg as a hard ceiling, and aim well under it. The data does not show a safe floor, so lower is the direction that costs you least.
- Count everything. Tea, chocolate, cola and the second cup you forgot all land on the same pile.
- Assume any coffee-shop drink is stronger than the label in your head. A double shot can be the whole day's limit on its own.
- If you are on several cups a day, do not stop dead. Withdrawal peaks at 20 to 51 hours and brings a headache you do not need right now. Step down over a week or two instead, the same way we lay out in the withdrawal timeline.
If you are trying
The CARE study's own recommendation was to bring intake down before conception. Most miscarriages happen in the first trimester, often before a woman knows she is pregnant. If pregnancy is the plan, the taper belongs before it.
If you are breastfeeding
The newborn's four-day half-life applies. Small amounts pass into milk and the baby cannot clear them, so they stack. Keeping intake low and having it straight after a feed is the conventional advice, and it lines up with the pharmacology.
If you are the partner
Going without alongside her is the single most useful thing you can do. Nobody manages a taper well while someone opposite them is on a second flat white.
The Uncaffed 30-day kit is built for that job. Reset steps you down by capsule count, from 160 mg a day to nothing over 20 days, so you never have to guess a brew. Flow runs alongside with no added caffeine for the full 30 days, carrying magnesium, which contributes to the reduction of tiredness and fatigue, and pantothenic acid, which contributes to normal mental performance. It is not for use during pregnancy or breastfeeding, and it is not for under-18s, so it is for the partner, or for the planning stage before a pregnancy.
Preorders are open. Kits ship by 30 October. How the step sizes were chosen is on the science page.
One more thing on the calculator. It models a non-pregnant adult at a 6-hour half-life. If you are in your third trimester, multiply every duration it gives you by about two and a half, and remember that a newborn's clock runs sixteen times slower than the calculator's.
The point
The 200 mg limit was set by what the studies could prove in 2008. Since then the cohorts have got bigger, the measurements have moved from questionnaires to blood tests, and the effects have shown up at lower and lower doses. Nobody has found the floor, and the likeliest reason is that there is not one. Every cup a pregnant woman drinks takes her about 15 hours to halve and her baby about four days.
