A due date, or estimated date of delivery, is counted from the first day of the last menstrual period (LMP), in one of two common ways: add 280 days, or add nine calendar months and seven days. They are often treated as one calculation. We ran both over every start date in one four-year leap cycle, 1 January 2021 to 31 December 2024: 1,461 dates. The calendar rule landed on the same day as 280 days on 642 of them (43.9%). On the other 819 it fell 1 to 3 days later, and never earlier. Where a date does not exist in the target month, we moved it to that month's last day.
Grünebaum, Dudenhausen and Chervenak measured this drift first, over 78,163 start dates, in a Research Square preprint posted on 8 July 2026 and not peer-reviewed; they call a gap of this size "clinically trivial". The date is an estimate either way. Naegele himself wrote that the day of birth can be foretold "nicht genau, sondern immer nur beiläufig": never exactly, only approximately.
A Latin passage quoted in 1812 dates pregnancy from a week after the last period
Franz Naegele's 1812 book Erfahrungen und Abhandlungen is often cited as the source of his rule. On pages 280 and 281 (page images), the passage concerns when a woman is most able to conceive. He cites the usual reckoning from the last period as evidence about conception, without offering a dating rule.
The counting words are a Latin quotation from Haller's edition of Boerhaave's lectures. It says 99 births in 100 happen "nono mense", in the ninth month, counting from a week after the last period. That is a month-wide window, resting on unnamed observations in France. Haller added his own notes, so whether these words are Boerhaave's or Haller's is unsettled. On page 282 Naegele quotes a line he attributes to a Hippocratic text, which we did not check, counting nine months from the menses or from conception. Pages 280 and 281 mention neither 280 days nor 40 weeks.
Textbook instructions for Baden midwives time the birth three months back and a week on
His own rule is in his textbook for Baden midwives, read here in its 1833 copy. Page 81 says to count from the day of the last period "3 ganze Monate zurück", three whole months back, then add seven days; no calendar needed, he says. His example runs from 10 June back to 10 March, then on to 17 March. By our count that is 280 days when the next February has 28 days, and 281 when it has 29. "Nine months and seven days" is a modern restatement, and his sentence does not say whether the first or last day of bleeding is meant.
Page 54 gives pregnancy as nine calendar months, or more exactly ten lunar months of 28 days, which is 40 weeks. That lunar month is a fixed 28-day count, unrelated to the months of the Chinese calendar.
The calendar shortcut postdates the 280-day count on most start dates
Our sweep of the 1,461 LMP dates, month ends clamped:
| Calendar rule against 280 days | LMP dates | Share |
|---|---|---|
| Same day | 642 | 43.9% |
| 1 day later | 398 | 27.2% |
| 2 days later | 308 | 21.1% |
| 3 days later | 113 | 7.7% |
Nine months from March or April skip February, so they run two days long. May is worst: 113 of the cycle's 124 May dates come out three days later, because the window takes in July, August, December and January. Only 29 to 31 May escape, pulled back by the clamp. A 29 February inside the window adds one more day.
Other conventions shift the counts slightly: rolling 30 February into March the JavaScript way, and counting back before adding a year, puts three late-May 2023 dates four days out. "Never earlier" held under all seven variants we tried. Under the clamp our script reproduces every count in the preprint's table; the formula its methods print, run as JavaScript, does not quite.
US obstetric convention schedules the due date 280 days after the last period
The American College of Obstetricians and Gynecologists (ACOG), in Committee Opinion 700 (2017, reaffirmed 2025), states in full: "By convention, the EDD is 280 days after the first day of the LMP." That assumes a 28-day cycle with ovulation on day 14, and ACOG reports that about half of women recall their LMP accurately. Our fertile window guide covers ovulation timing.
Our due date calculator counts 280 days, as ACOG does. Until 16 October 2026 its text said the calendar shortcut "gives the same answer" and called LMP plus 280 days Naegele's rule of 1812. An out-of-range year left a stale result with no message, and a future start of 1 March 2027 gave "Week 0" with 416 days remaining. The revision made that day shows the calendar rule's date beside the 280-day one and refuses a future date or one more than 44 weeks back. A start of 1 May 2026 with the default 28-day cycle gives 5 February 2027, with 8 February, three days later, on the comparison line.
Births fall within a week of the estimated date about half the time
Mongelli, Wilcox and Gardosi studied 34,249 single pregnancies with "certain" period dates in England's East Midlands (their 1996 abstract). Births within 7 days of the period-based date made up 49.5%, against 55.2% when the pregnancy was dated by ultrasound. Within 10 days the shares were 64.1% and 70.3%. The often quoted "4% born on the due date" comes from a later paper's introduction citing Mongelli; his abstract does not contain it.
That later paper, Jukic and colleagues' 2013 study, followed 125 births in North Carolina from 1982 to 1985. Counted from the period, the median pregnancy was 285 days, the middle half falling between 277 and 290. Counted from measured ovulation, the median was 268 days, and lengths still spanned 37 days with six preterm births set aside. Smith's 2001 study of 1,514 women whose period and early scan dates agreed within a day found a median of 283 days to delivery that was not induced or planned.
Early scans redate a pregnancy by thresholds that widen as it advances
A dating scan measures the fetus itself, early on by its crown–rump length. ACOG puts that within 5 to 7 days up to 13 weeks and 6 days, written 13 6/7. It advises changing the date if a scan before 9 weeks differs from the period date by more than 5 days, or one before 14 weeks by more than 7. Accuracy then loosens to 7 to 10 days at 14 to 21 weeks and 21 to 30 days from 28 weeks. A pregnancy with no scan before 22 weeks it calls "suboptimally dated".
NICE in England, in guideline NG201 (2021), recommends offering a scan between 11+2 and 14+1 weeks to "determine gestational age". KKH, a Singapore public hospital, lists a dating scan to estimate the due date in weeks 8 to 12 of its published pathway, with optional first trimester screening in weeks 11 to 14.
View author profile → · Editorial policy
