Guessing "boy" every time does at least as well as the Chinese gender chart. Slightly more boys than girls are born everywhere, about 105 for every 100 girls on the United Nations' modelled figures for 2024, so always saying "boy" would be right about 51.4% of the time in Sweden or Singapore and about 52.5% in China or Viet Nam. That is our arithmetic, not a study. A chart whose answers have nothing to do with the baby's sex cannot do better than that, and the largest test of the Chinese chart found its answers had nothing to do with the baby's sex.
The chart, known in Chinese as 清宫表 or 清宫图 (the "Qing palace chart"), reads a boy or a girl off a grid of the mother's age and the lunar month of conception. Pages selling the 2027 "authentic edition" (正版) were already online in September 2026, and some Indonesian and Malay pages advertise it as 90% or 93% accurate. It is not medical advice and no part of this guide is. In China, identifying a fetus's sex by ultrasound or other technical means without a medical reason is prohibited (Population and Family Planning Law, Art. 39); in Viet Nam, the Population Law of 2025 prohibits selecting a fetus's sex and revealing it, except where the health minister allows it to diagnose and treat sex-linked disease, from 1 July 2026 (Art. 6.3).
Against a birth register
The biggest test used 2,840,755 single births in Sweden between 1973 and 2006. The researchers worked out each mother's lunar age and month of conception, read the chart, and compared its prediction with the sex of the baby. Agreement was measured with Cohen's kappa, where 0 means no better than chance: it came out at 0.0002, with a 95% confidence interval from −0.0009 to 0.0014. The chart was "no better at predicting the sex of a baby than tossing a coin", the authors concluded, and they advised "against painting the nursery based on this method's result". We read the abstract; the full paper would not open for us, so we cannot say which version of the chart they used.
A smaller test at a hospital in Jinan in 2017, 107 births, found the chart's accuracy below 50%, and reported a higher figure for one group, second babies, at 56.7%. At that size, chance alone produces accuracy anywhere between about 40% and 60% in 95 of 100 runs, and a subgroup of that sample swings wider still, so the second-baby figure is what noise looks like. Our guide to how big a test needs to be explains why a hundred cases cannot settle a question like this.
Against the other copies
If the chart came from a single ancient document, every copy would be the same. We compared three widely copied versions cell by cell across all 336 cells, 28 ages by 12 months. Two of them, on cn.5d8z.com and yourchineseastrology.com, agree on 330 cells. The third differs from each of those on 38 and 42 cells.
Our own Chinese gender chart tool was further off than any of them. Until September 2026 its grid disagreed with the three copies on 140 to 154 cells, and two of its rows, for mothers aged 18 and 21, were the exact reverse of the common chart. Its own default example answered "boy" where all three copies say "girl". It now uses the chart the two closest copies share, shows both readings on the six cells where they differ, and no longer shows a whole year's row for one age, which turned it into a planner.
Against the calendar
The chart needs the mother's age in the traditional count, 虚岁, in which a baby is one at birth and gains a year at each Lunar New Year, and the lunar month of conception. Neither has a single agreed rule. Some pages count 虚岁 by lunar New Years, some take the Western age and add one, some add nine months. Leap months, which some lunar years have, are handled three different ways, and some pages use the solar-term months of the Chinese almanac instead.
We ran about half a million pairs of mother's birth date and conception date through the old tool's own code. Switching from the lunar-year count to "Western age plus one" changed the age for 49% of the date pairs we tested and the chart's answer for 33.5% of them. Our guide to why your Chinese age is higher explains the two counts, and why solar-term dates move covers the other calendar.
Against its own story
The origin stories do not agree with each other either. One says the chart lay in a royal tomb near Beijing for "nearly 700 years", which would date it to around the 1300s, more than three centuries before the Qing dynasty that gives the chart its name. Another says the Qing court's eunuchs kept it until foreign troops carried it to England in 1900. A third says a Chinese scholar copied it back from Austria. A fourth places it in an "Institute of Science in Beijing". None of them names a museum, an archive, a catalogue number or a date of publication, and we could find none. A chart said to have served the imperial court for centuries would be a famous object in whatever collection held it. None of the stories says which collection that is.
View author profile → · Editorial policy
