Reference · 14 methods · sources checked 2026

Old wives' tales for gender prediction

Every list of these you have ever read gives you the tales and stops. Nobody says where any of them came from, and nobody checks. So we did both. Below are 14 methods for guessing a baby's sex before the scan, each traced back as far as it can honestly be traced, and each set against whatever published evidence exists.

The short version: 2 survive, and neither of the two is something you can do at your kitchen table. 7 were tested by somebody and came back at chance. 5 have never been tested at all, mostly because there was never a claim specific enough to test. One of them will make you light-headed and should not be attempted. The table says which is which.

The table

Holds up survived a published test · Coin flip was tested and came back at chance · Never tested nobody has ever measured it

Method	Verdict	Earliest traceable source	What the evidence says
Nub theory At the 12-week scan, the angle of the genital tubercle against the spine calls the sex: level means girl, angled up means boy.	Holds up	Efrat, Nicolaides and colleagues, Ultrasound in Obstetrics & Gynecology, 1999: an actual clinical paper, with a named method and a sample.	About 70% correct before 12 weeks, rising to the high nineties at 13 weeks in the follow-up work. The catch is that this is a sonographer's measurement taken in a true sagittal plane; squinting at a phone photo of a scan printout is a different, untested activity. Source
Severe morning sickness The worse the sickness, the more likely it is a girl.	Holds up	Folk everywhere, but first properly measured by Askling and colleagues in The Lancet, 11 December 1999, on Swedish national registry data.	A real but small excess of girls among women hospitalized for hyperemesis gravidarum. Two honest caveats: it shifts the odds by a few points rather than telling you anything, and it applies to hospital-grade sickness, not to ordinary nausea. Note that nobody sells this one. Source
Ramzi theory On a six-week scan, a placenta on the right means a boy, on the left a girl.	Coin flip	Saam Ramzi Ismail, self-published in 2011 and never peer reviewed. He reported 5,376 participants and accuracy above 97%; the write-up was later replaced online with a note that the method has not been confirmed elsewhere.	An independent study of 277 pregnancies in Ultrasound in Obstetrics & Gynecology found no relationship between placental location and fetal sex. No one has replicated the original, and no proposed mechanism explains why sex would move a placenta. Source
Skull theory The shape of the skull on the 12-week scan: squarer for boys, rounder for girls.	Never tested	Pregnancy forums, mid-2010s. No originator, no paper, no stated measurement.	Never tested. It is sold by the same services that sell nub readings, which is worth knowing when the two are quoted side by side at the same confidence.
The 140 heartbeat rule A fetal heart rate above 140 beats per minute means a girl, below means a boy.	Coin flip	No traceable originator. In circulation in pregnancy books by the 1990s, attributed to nobody.	McKenna and colleagues compared 477 first-trimester heart rates: 154.9 bpm for boys against 151.7 for girls, no significant difference. A 2023 systematic review reached the same verdict, and 140 does not fall where the rule needs it to. Source
Shettles method Time intercourse near ovulation for a boy, several days before for a girl.	Coin flip	Landrum Shettles, Your Baby's Sex: Now You Can Choose, 1970, written with David Rorvik.	Wilcox, Weinberg and Baird tracked 625 cycles for the New England Journal of Medicine in 1995 and found timing had no bearing on the sex of the baby. Worth flagging that this is a selection method, not a prediction method; it is on this list because it is sold alongside them. Source
The Qing palace chart A 336-cell grid reading the mother's lunar age against the lunar month of conception.	Coin flip	The tomb, the institute, the seven centuries: we went through all of it line by line.	The full provenance trace and the accuracy evidence live on the chart's own page, along with the cells on which published copies contradict each other. Read the full trace
Mayan calendar chart Add the mother's age to the year of conception; both even or both odd means a girl.	Never tested	Nothing Maya about it. No codex, no scholar, no date. The name is borrowed and the rule is a modern web invention.	This one you can settle without a study. A parity rule over two roughly arbitrary integers splits its answers close to evenly by construction, so it is a coin flip for the same reason a coin is.
Drano test Unsafe Mix urine with drain cleaner and read the color that comes up.	Coin flip	Circulating by the late 1970s. The earliest formal check we can find is a University of Wyoming trial on 100 pregnant women in the early 1980s.	No better than chance, and individual women got different colors in different months of the same pregnancy. Separately: lye and urine together give off fumes, the manufacturer has never sold the product for this, and there is no version of this test worth running. Please skip it.
Baking soda test Drop urine into baking soda; fizzing means a boy.	Never tested	No origin at all. It appears in the 2010s as a Drano test with the dangerous ingredient taken out.	Never measured. Baking soda fizzes on contact with acid, and urine pH moves with diet, hydration and time of day, all of which vary far more than anything fetal.
Ring on a string Suspend a ring over the bump: circles mean one sex, a straight swing the other.	Never tested	Older than its pregnancy use by centuries: it is pendulum divination, repurposed. Which sex gets the circle flips between tellings.	Never tested for this, though the mechanism is well described elsewhere: the ideomotor effect, in which the hand supplies tiny movements the holder does not notice making.
Carrying high or low Low and out front means a boy; high and spread wide means a girl.	Coin flip	Folk, worldwide, unattributable.	Perry, DiPietro and Costigan at Johns Hopkins put pregnancy folklore to the test on 104 women for the journal Birth in 1999. Predictions made on folklore grounds (belly shape, sickness) did not beat chance. Predictions made on dreams and feelings did better, which is the funniest result in this entire table. Source
Sweet or salty cravings Sweet cravings mean a girl, salty or sour mean a boy.	Coin flip	Folk, unattributable, and inverted in some regions.	Falls inside the folklore group in the same Johns Hopkins test, which found no systematic relationship between any of these markers and the actual sex. Source
The linea nigra, skin and hair The dark line's path tells the sex; a girl "steals the mother's beauty."	Never tested	Folk, unattributable.	Never tested. The line is pigmentation driven by pregnancy hormones and appears in pregnancies of both sexes; skin and hair changes have the same cause.

Three things the table shows that the listicles don't

The two survivors are both clinical. Nub theory works because a sonographer is measuring an angle in a properly aligned image, not because there is folk wisdom in it; take the same idea out of the clinic and into a phone photo and you have removed the part that did the work. Severe sickness leans toward girls because hyperemesis is a hospital diagnosis with a hormonal profile behind it, not because nausea is feminine. Both of the winners, in other words, stop working the moment you try to do them yourself. That is the actual finding here, and no listicle states it.

Nobody cites anybody. We tried to find a first appearance for each method, and for most of them the trail ends at "people say." That is fine, because folklore does not keep receipts, and nobody expects a footnote on a ring and a piece of string. What is less fine is the confidence. The methods with no source at all are quoted with exactly the same certainty as the ones with a named researcher and a sample size, and usually in the same paragraph of the same article. Two of these methods have real originators, Landrum Shettles in 1970 and Saam Ramzi Ismail in 2011. Both got tested precisely because you could find them and check. That is the argument for sourcing, in miniature.

The funniest result belongs to the mothers. When the Johns Hopkins team asked 104 women how they had arrived at their guess, the ones reasoning from folklore, belly shape and sickness and cravings, did no better than a coin. The ones who said they had simply dreamed it, or had a feeling, did better. Take that however you like. We would only note that after 14 methods, 2 of which hold up and neither of which you can run at home, a hunch is not obviously the least scientific instrument in the room.

How we judged these

Earliest traceable source means the earliest named person, publication and year we could actually reach, not the earliest claimed. Where a method is genuinely unattributable we say so rather than inventing a lineage, and where a famous origin story turns out not to survive checking, we say that too. Never tested means no study exists, which is not the same as disproven; it usually means the claim was never stated precisely enough to test. Coin flip is reserved for methods somebody actually measured and found at chance.

None of this is medical advice, and none of it changes anything about your pregnancy. For the real answer, cell-free DNA screening reports fetal sex from about 10 weeks and the anatomy scan at 18 to 22 weeks is the standard check. Ask your midwife or doctor. Everything on this page is a game, and it is a better game when you know which parts of it are made up.

Questions people ask

Which gender old wives' tale is the most accurate?

Of the folk methods, none. The only two entries on our list that survive a published test are not really old wives' tales at all: nub theory, which is a sonographer's measurement taken at the 12-week scan, and severe morning sickness, where hospital-grade hyperemesis carries a small statistical lean toward girls. Everything you can do at home (the ring, the baking soda, the belly shape, the heartbeat rule) has either been tested and failed, or has never been tested at all.

How accurate are old wives' tales for gender prediction?

About 50 percent, which is what you get from guessing. A Johns Hopkins team tested pregnancy folklore directly on 104 women in 1999 and found that predictions made on folkloric grounds did not beat chance. Curiously, the women who went on a dream or a hunch did better than the ones who went on the belly shape.

Are any of these tests dangerous?

One is. The Drano test asks you to mix urine with a lye-based drain cleaner, which releases fumes and has no upside, since it was tested in the early 1980s and found no better than chance. There is no safe way to run it and nothing to learn from it. The rest are harmless, if you count a small amount of baking soda and some disappointment as harmless.

Why do so many of these have no source?

Because folklore does not keep receipts, and the internet did not start. When we tried to trace each method to a first appearance, most of them stopped at "people say." A few have real originators: Landrum Shettles in 1970, Saam Ramzi Ismail in 2011. Those are the ones you can actually check, which is how they came to be tested and found wanting.

When can I actually find out my baby's sex?

Cell-free DNA screening (NIPT) can report fetal sex from about 10 weeks, and the routine anatomy ultrasound at 18 to 22 weeks is the standard check. A well-taken 12 to 13 week scan read for the nub angle is genuinely informative but is a clinical measurement, not something to read off a photo. Ask your midwife or doctor.

One row, in full

The Qing palace chart, taken apart

The one method on this list with an imperial legend attached. We traced the tomb, the institute and the seven centuries, and none of them are there.

Once you know

A Chinese name for the baby

The guessing games are a coin flip. A name chosen for the child's birth year and element is the part that keeps.