The clearest evidence on how much a due date can be off comes from a study that could measure pregnancy length precisely, because it tracked confirmed ovulation dates instead of guessing from a last period. Jukic and colleagues followed 125 women trying to conceive naturally, timed ovulation directly, and found a median pregnancy length of 268 days from ovulation, close to but not identical to what Naegele's rule predicts, with a normal range spanning roughly five weeks even after excluding preterm births (Jukic et al., Human Reproduction, 2013).
A due date built on Naegele's rule assumes ovulation on day 14 of a 28-day cycle, described in full on the due date calculation guide. It is a useful anchor for scheduling prenatal care, not a prediction of a specific delivery day, and the research on natural variation backs that up directly.
In the Jukic study, even pregnancies that reached full term and had no complications varied in length by close to five weeks from shortest to longest. That is a wider spread than most people expect from a single calculated date, and it is the reason a due date functions better as the center of a window than as a target.
ACOG Committee Opinion No. 700 puts first-trimester crown-rump length dating at an accuracy of about plus or minus 5 to 7 days for establishing gestational age (ACOG, 2017). That narrows the chance that the calculated date itself is wrong. It does nothing to narrow the separate, biological fact that pregnancies of identical, correctly measured gestational age still deliver on different days within a normal range, which is what the Jukic data describes.
Population-level data generally shows first pregnancies trending a little longer than later ones, on average, but the spread within either group swamps that difference for any individual pregnancy. The practical takeaway is the same either way: a due date tells you roughly where the middle of the distribution sits, not which day you personally will land on.
Full term under ACOG's own categories spans 39 weeks 0 days through 40 weeks 6 days, described in the full-term guide, and most providers do not treat "past the due date" as urgent on its own until well into that window. If your due date and an ultrasound date disagree by more than the threshold in ACOG's redating table, that is worth asking your provider about directly; if they simply differ from what you expected, that is closer to the range this page describes.
See your estimated due date and current week, built from your own dates.
Jukic and colleagues followed 125 women trying to conceive naturally, recorded the date of ovulation directly rather than estimating it from a last period, and measured how long each pregnancy actually lasted from that confirmed date to delivery.
Population data generally shows first pregnancies running slightly longer on average than later ones, though the spread within either group is wide enough that due date alone predicts very little about any one birth.
It narrows the dating error, not the biological variation. A first-trimester crown-rump length measurement is accurate to within about 5 to 7 days for establishing gestational age, per ACOG, but it cannot predict which day within the normal range labor will actually start.
Yes. It sets the window providers use to time screening tests, monitor growth, and decide when a pregnancy is starting to run late enough to discuss induction. Its value is in structuring care, not in predicting a birthday.