How pregnancy due dates are calculated (and why they're usually wrong)

The standard pregnancy due date is calculated using Naegele’s rule: take the first day of the last menstrual period (LMP), subtract three months, add seven days. The result is 280 days — exactly 40 weeks — from LMP. It’s simple, memorable, and based on assumptions that rarely match reality. Only about 4% of births happen on the predicted day. About half happen within ±5 days. The rest sit outside even that window.

Where 280 days comes from

Franz Karl Naegele published the rule in 1812. He assumed:

• A regular 28-day menstrual cycle.
• Ovulation on day 14 of the cycle.
• Conception immediately at ovulation.
• Gestation of 266 days from conception, which equals 280 days from LMP.

Two centuries later we know each of those assumptions has a meaningful error bar. Cycle lengths vary; ovulation timing varies within a single cycle; the conception window can span 6 days; gestation duration itself has a spread of ±2 weeks even after controlling for the LMP date.

What the data shows

A 2013 study at Cedars-Sinai (Jukic et al., Human Reproduction) measured gestation directly using hormone-confirmed conception dates rather than LMP. The median gestation was 268 days from conception — about 2 days longer than Naegele assumed. The 90% confidence interval was ±17 days. That means a baby arriving any time between 37 weeks and 42 weeks from LMP is statistically unremarkable.

The CDC’s National Vital Statistics System data on US births shows:

Why the due date is still useful

Even though only 4% of babies are born on the date, the date is the reference point for almost every prenatal clinical decision: ultrasound timing, fetal anatomy scans, glucose-tolerance testing, group B strep screening, induction policy if gestation exceeds 41 weeks. The number is a clinical anchor, not a prediction.

Modern obstetrics typically refines the LMP-based estimate with a first-trimester ultrasound (crown-rump length is accurate to ±5 days at 8-13 weeks). If the ultrasound disagrees with LMP by more than 7 days, ACOG (American College of Obstetricians and Gynecologists) recommends switching to the ultrasound date.

Why first pregnancies run later

Multiple studies (Jukic 2013, Mongelli 1996, Smith 2001) agree: first pregnancies last about 5 days longer than subsequent ones, on average. The biological mechanism is unclear — possibly cervical-tissue conditioning, possibly hormonal-feedback differences. The practical implication: first-time mothers should plan for a delivery window skewed past 40 weeks rather than centered on it.

What to ask instead

The honest version of the question “when’s the baby coming?” is “what’s the 90% delivery window?” — typically 37+0 to 41+6 weeks from LMP. That’s a 35-day spread. Maternity leave planning, relative travel arrangements, and finished-nursery deadlines should be tied to the start of that window, not the center.

Use our due date calculator to get the Naegele estimate plus the 90% window from your LMP date. For LMP-from-conception conversion (helpful with IVF or known conception dates), the calculator does the offset for you.

Edge cases the rule doesn’t handle

• Irregular cycles. Naegele assumes 28 days; cycles routinely span 21-35. A 35-day cycle ovulates ~7 days later than the rule assumes, so the due date should be pushed back by 7 days. Many calculators include a cycle-length adjustment.
• IVF.Conception date is known precisely. Standard adjustment: due date = transfer date + 263 days (for a day-5 blastocyst transfer). LMP-based calculators don’t apply here.
• Twins. Twin gestations are typically ~36-37 weeks rather than ~40. The Naegele-derived 40-week due date is essentially never reached by a twin pregnancy.

The pragmatic takeaway

Treat the due date as the midpoint of a 5-week window, not a prediction. First pregnancies skew late; cycle-adjusted due dates are more accurate; ultrasound-confirmed dates are more accurate still. The exact date matters far less than the window, and the window is wider than most first-time parents are told.

Worked example

First pregnancy. LMP = 2026-01-08. Cycle length 32 days (longer than the Naegele-assumed 28).

• Raw Naegele: LMP − 3 months + 7 days = 2025-10-08 + 7 = 2026-10-15. Or equivalently LMP + 280 days.
• Cycle adjustment: ovulation in a 32-day cycle lands at day ~18, not day 14. Add 4 days. Adjusted due date: 2026-10-19.
• First-pregnancy adjustment: Jukic data suggests first pregnancies last ~5 days longer than the population median. Practical midpoint: 2026-10-24.
• 90% delivery window: 37+0 weeks (2026-09-24) through 41+6 weeks (2026-10-30). A 36-day spread.
• If a first-trimester ultrasound at 9 weeksgives a crown-rump length corresponding to gestational age 9w 5d (instead of the LMP-implied 9w 0d), ACOG recommends switching the dating: new EDD = 2026-10-19 (5-day forward shift). Ultrasound at <14 weeks is more accurate than LMP and supersedes the cycle math.

Common mistakes

• Treating the EDD as a deadline.ACOG considers 39+0 to 40+6 weeks “full term.” Earlier (37-38+6) is “early term” with measurably worse neonatal outcomes; later (41+0 to 41+6) is “late term.” Induction before 39+0 for non-medical reasons is contraindicated by ACOG.
• Using a calculator on an IVF pregnancy.LMP-based calculators add 280 days; IVF requires transfer-date arithmetic (day-5 blastocyst + 263 days, day-3 cleavage + 265 days). The errors are ~5-14 days.
• Ignoring a cycle-length discrepancy. A 35-day cycle ovulates ~7 days late, so Naegele overestimates gestational age by 7 days. This matters for first-trimester screening timing.
• Switching between “weeks pregnant” conventions.Most obstetric counting is from LMP (so “8 weeks pregnant” is ~6 weeks since conception). Fertility-clinic conventions sometimes count from conception. Always specify which.
• Recalculating after each ultrasound. ACOG recommends the EDD be set once, ideally at the earliest first-trimester ultrasound, and not changed subsequently. Later ultrasounds have wider error bars (±2 weeks at 28 weeks, ±3 weeks at 36 weeks).

When the rule shouldn’t be used at all

• Unknown or unreliable LMP. First-trimester ultrasound is the standard of care; LMP-based calculation is replaced entirely.
• Recent hormonal contraception.Post-pill cycles can be irregular for 3-6 months; the “LMP” may not represent a true menstrual cycle.
• Breastfeeding-related conception.Lactational amenorrhea before the first postpartum period means there’s no LMP to date from.
• Higher-order multiples. Triplet and higher gestations typically deliver at 32-34 weeks; the 40-week framework is irrelevant.

Sources: Jukic AM et al., Human Reproduction 2013; ACOG Committee Opinion 700 (Methods for Estimating the Due Date, 2017); WHO antenatal care recommendations 2016; CDC National Vital Statistics Reports Vol. 72, No. 1, 2023.