All topics › Pregnancy preparing for birth
How much caffeine is safe in pregnancy, and what do we actually know about miscarriage, stillbirth, and fetal growth?
Most people in the UK drink caffeine daily, and most pregnancies begin with a coffee habit. Guidelines say "no more than 200 mg a day" — roughly two mugs of instant coffee. But what evidence sits behind that number, and does it actually protect against miscarriage or poor fetal growth?
The UK guidance (NHS) is to limit caffeine to 200 mg a day in pregnancy [1]. The American College of Obstetricians and Gynecologists (ACOG) says moderate intake below 200 mg/day "does not appear to be a major contributing factor" to miscarriage or preterm birth, but adds that the relationship to growth restriction is undetermined and that no final conclusion can be reached on high intake and miscarriage [2].
Honestly? The evidence behind the number is thinner than the guidance sounds. Nearly everything linking caffeine to miscarriage, stillbirth, or smaller babies comes from observational studies, where the biggest confounder is nausea: healthy pregnancies cause nausea, and nausea makes people stop drinking coffee — so coffee drinkers can look like they're at higher risk even if the coffee isn't the cause [3][4]. One randomized trial (of switching to decaf mid-pregnancy) found no effect on birthweight or gestation [5]. The 200 mg ceiling is a sensible precaution, not a proven safe/unsafe boundary.
NHS guidance sets the limit at 200 mg of caffeine per day through pregnancy [1]. The approximate caffeine content of common drinks (NHS regional pages): a mug of instant coffee ~100 mg, a mug of tea ~75 mg, a can of cola ~40 mg, and a 250 ml energy drink up to 80 mg [1]. Filter coffee is typically stronger than instant, but strength varies too much by brew to give a single number — count it as more than instant. Two mugs of instant coffee plus a cup of tea already takes you over the line. Note that these are estimates — actual coffee strength varies a lot by how it's made.
The largest synthesis (Greenwood et al., 2014: 60 publications from 53 observational studies) found that each additional 100 mg of caffeine per day was associated with a 14% higher relative risk of miscarriage (risk ratio 1.14, 95% CI 1.10–1.19) [3]. That sounds alarming — but it is association, not proof. The same review found substantial variation between studies, signs of publication bias, and incomplete adjustment for smoking and obstetric history [3].
The nausea problem is the key reason to be cautious. In Weng et al.'s 2008 prospective cohort (1,063 women; 172 miscarriages), intake was grouped as none, under 200 mg/day, and 200 mg/day or more. After adjustment — including for nausea and vomiting — the under-200 group had an adjusted hazard ratio of 1.42 and the 200-plus group 2.23 (95% CI 1.34–3.69), versus no caffeine [4]. Weng's team reported that adjusting for nausea and vomiting did not materially change the association — but self-reported nausea is a crude measure, and the deeper worry (that healthy pregnancies cause both nausea and coffee avoidance) can't be fully adjusted away. And the evidence isn't one-sided: ACOG's committee opinion notes a large cohort study (Savitz et al.) that found no association between caffeine and miscarriage at all [2].
Greenwood's 14%-per-100-mg figure is relative only: miscarriage baselines vary too much by maternal age and how early pregnancies are counted for any single absolute conversion to be valid.
Greenwood's review also found a dose-response association with stillbirth: about 19% higher risk per additional 100 mg/day [3]. Stillbirth is rare in the UK (about 3.5 per 1,000 total births in 2021 [7]), so even if that association were causal, an extra 100 mg/day would move the absolute rate from ~3.5 to ~4.2 per 1,000 — less than one extra stillbirth per 1,000 pregnancies. The review's authors reported the association but the finding rested on relatively few studies [3].
For growth, Chen et al.'s 2014 meta-analysis found that moderate caffeine intake (150–349 mg/day) was associated with babies born about 33 g lighter (95% CI 4–63 g) than babies of women with no or very low intake [8]. Thirty-three grams is roughly 1% of a typical newborn's weight — the clinical significance is debatable.
Bech et al. (2007) randomized pregnant coffee-drinkers to caffeinated or decaffeinated instant coffee from mid-pregnancy. The decaf group cut their caffeine intake substantially — and the trial found no significant differences in birthweight or length of gestation [5]. This is the only experimental evidence on caffeine in pregnancy. The honest limitation: the intervention started in mid-pregnancy, so it cannot answer whether caffeine matters in early pregnancy, when miscarriage risk is the question [5].
Sleep is the parental outcome caffeine most clearly affects. ACOG's own patient guidance notes that caffeine can interfere with sleep, contribute to nausea and light-headedness, and increase urination [9] — and pregnancy already does all of those to you. If cutting back helps you sleep, that's a benefit to you, not just a theoretical benefit to the baby. Nobody has run a trial on whether caffeine reduction improves pregnant parents' sleep, mood, or day-to-day functioning — that's a gap in the evidence — but it's the most tangible reason to care.
The guilt trap is worth naming directly. Most early miscarriages are caused by chromosomal abnormalities and could not have been prevented [6]. If you drank coffee before you knew you were pregnant, or struggled to cut down, the evidence does not establish that your coffee caused harm — the associations are small, confounded, and the one randomized trial found no measurable difference. Cutting back now is a reasonable precaution, not a confession.
Practically: if you love your coffee, two mugs of instant a day is roughly the guidance ceiling; filter coffee gets you there faster. Decaf is the straightforward swap (the trial used instant decaf from mid-pregnancy with no measured downside [5]). Tea counts. Energy drinks count too. Chocolate's contribution is small.
| Exposure / comparison | Outcome | Effect in absolute terms | Evidence rating |
|---|---|---|---|
| Each extra 100 mg/day | Miscarriage | ~14% higher relative risk per 100 mg/day [3] (observational; causality unproven; relative only — no valid single absolute baseline) | C |
| ≥200 mg/day vs none (Weng 2008) | Miscarriage | Adjusted hazard ratio 2.23 (1.34–3.69) [4]; cohort overall 172/1,063 ≈ 162 per 1,000 | C |
| Each extra 100 mg/day | Stillbirth | ~19% higher relative risk [3]; illustratively, ~3.5 → ~4.2 per 1,000 at the UK rate [3][7] (observational) | C |
| Moderate intake 150–349 mg/day vs none | Birth weight | About 33 g lower (95% CI 4–63 g) [8] | C |
| Switching to decaf (RCT, from mid-pregnancy) | Birthweight / gestation | No significant difference in either outcome [5] | B (for the null; mid-pregnancy onwards only) |