32,329,859 babies · United States 2016–2024 · observed birth weights, not predictions
Enter the week of pregnancy your baby was born and your baby's birth weight. The tool shows how that weight compares with babies born in the same week in the United States between 2016 and 2024. It does not say whether a weight is normal; there is barely such a thing. It says where your baby's weight sits among millions of others.
You will see the 5th, 10th, 50th, 90th and 95th percentiles for that week, a growth chart from 22 to 42 weeks with your baby marked on it, and your baby's own percentile. A percentile is a rank: if your baby is at the 30th percentile, about 30 out of every 100 babies born in that week weighed less, and about 70 weighed more.
You can then narrow the comparison to babies more like yours: the baby's sex, whether this is a first baby, the mother's age, height and body mass index, diabetes, high blood pressure, smoking during pregnancy, and the race and ethnicity of the mother and of the father. The chart, the table and the percentile all recalculate, and the tool always tells you how many births are in the comparison.
Every number is a direct count from the birth certificates of every singleton baby born in the United States from 2016 to 2024, collected by the National Center for Health Statistics. Nothing is modelled, fitted, smoothed or estimated. Choose a week and a set of characteristics, and the tool counts the matching babies and ranks them by weight.
Doctors have argued for thirty years about whether a baby's weight should be judged against all babies or against babies of similar parents. A separate page explains the argument, what this tool does differently from every earlier chart, and what it cannot tell you.
Term cesarean rate, all births from 22 weeks, Robson groups and compare states use the same birth certificate data to look up cesarean rates. All of them open with every menu on Any and narrow only where you choose.
United States 2016–2024 · NVSS natality microdata · observed birth weights, not predictions
The middle weight is the median: half of the babies born that week weighed less and half weighed more. The usual range holds the middle 80 per cent of babies. Eight in ten babies fall inside it; one in ten is below and one in ten is above, and most of those are healthy too.
Weeks with fewer than 30 matching babies are left off the chart. Weeks with fewer than 100 are drawn with hollow markers and should be read with care. Each point is an exact percentile of the babies born in that week; the lines only connect the points.
Percentiles are computed from birth weights recorded in 25 gram steps and interpolated within the step, so each value is accurate to about 12 grams, or under half an ounce.
Rows in grey have fewer than 100 babies. Rows with fewer than 30 babies are shown with a count only.
Every publication, definition and exclusion behind the figures
Who is counted. Every live-born singleton baby in the United States from 2016 to 2024 whose mother lived in the United States, born at 22 to 42 completed weeks by the obstetric estimate of gestation (OEGest_Comb, positions 499–500), with a recorded birth weight of at least 500 grams (DBWT, positions 504–507). 32,329,859 babies. All birth settings are included, hospital, birth centre and home, because a reference should describe all babies born. Twins and higher multiples are excluded because they are lighter at every week and would pull every curve down.
Why the obstetric estimate. The birth certificate carries two measures of gestation: the clinician's obstetric estimate and the calculation from the last menstrual period. The obstetric estimate is the NCHS standard since 2014 and produces far fewer impossibly heavy "preterm" babies, which are almost always misdated term babies. Reference 4 shows how much those errors distort the preterm percentiles.
Why 500 grams. Below 500 grams the file mixes extremely preterm live births with births at the edge of viability that are reported very differently from state to state. The floor removes them. It is the same floor used by the other tools on this site.
Percentiles. Birth weights are counted in 25 gram bins and the percentile is interpolated within the bin, so each figure is accurate to about 12 grams. Nothing is smoothed across weeks: each week's percentiles come only from the babies born in that week. This is why a curve can look uneven where a comparison has few babies; the tool leaves a week off the chart below 30 babies and marks it as thin below 100.
A visible kink at 28 to 31 weeks. With every menu on Any, the 90th and 95th percentiles rise sharply from 29 to 30 weeks and then flatten to 31. That is the data, not a mistake in the tool: a small share of the babies recorded at 30 weeks are heavier than a real 30-week baby can be, most likely term or near-term babies whose week of gestation was misrecorded or rounded to 30. Reference 4 describes this problem and one way of trimming it. Nothing is trimmed here, because the tool reports what the certificates say; read the upper percentiles between 28 and 31 weeks with that in mind.
Your baby's percentile is the share of babies in the comparison, born in the same week, who weighed less than the weight you entered, again interpolated within the 25 gram bin.
Baby's sex is the infant sex item. First baby is the live birth order recode: 1 means no previous live birth. Mother's age is single years of age at delivery, in the same five bands used across this site. Mother's height is the reported height in inches, in five bands plus Unknown; it is the one maternal characteristic that the critics of customised charts accept as pure physiology, because babies of short women are smaller without being at higher risk (reference 7). Body mass index is the pre-pregnancy value the certificate computes from reported height and pre-pregnancy weight, in six bands plus Unknown.
Diabetes is Yes if the certificate records pre-pregnancy diabetes or gestational diabetes or both; No if both are recorded as absent. High blood pressure is built the same way from pre-pregnancy hypertension and gestational hypertension (eclampsia is not a separate menu). Smoking is the cigarette recode: any cigarettes in any trimester. Where an item is unknown or not stated, the baby stays in Any but is not counted under Yes or No.
Race and ethnicity of the mother and of the father come from the two combined race and Hispanic origin items. Hispanic of any race is one group; the others are non-Hispanic White, Black and Asian, and Other combines American Indian and Alaska Native, Native Hawaiian and Other Pacific Islander, and more than one race. The father's race and ethnicity is missing or not stated on about one certificate in six, most often when the father is not named. It is kept as Unknown or not stated so that Any loses nobody, and the count on screen shows how many births a specific choice keeps.
Every menu starts on Any. Each choice you add keeps only the babies who match, so the comparison gets more like your pregnancy and smaller at the same time. The counts are always shown. When a week falls below 30 matching babies, the tool says so instead of showing a percentile built on a handful of births.
A reference describes what happened: the weights of babies who were actually born. A standard describes what should happen: the weights of babies growing under good conditions. This tool is a reference. When a menu is set to a condition that changes growth, diabetes, high blood pressure or smoking, the percentiles describe babies whose growth was changed by that condition, which is why every customised chart since Gardosi's of 1992 leaves those conditions out of its model (references 8, 9) and why this tool keeps the comparison with all babies on screen whenever they are chosen.
Whether a baby's weight should be judged against similar parents at all is unsettled. Gardosi and colleagues hold that adjusting for the mother's height, weight, parity and ethnic group finds more of the babies at risk (reference 8). Hutcheon, Zhang, Kramer and colleagues hold that those characteristics explain differences between populations but barely predict an individual baby's weight, and that adjustment adds almost nothing to the prediction of death (references 7, 10). A Scottish study of 979,912 term births found that partial adjustment did not improve prediction (reference 11). This tool takes no side. It shows the actual distribution for whichever comparison the user chooses and states how many babies are in it.
It cannot say whether a baby is healthy. Birth weight percentile is one measurement, and the doctor or midwife reads it together with the reason the baby was born when it was, the baby's length and head size, the placenta, and how the baby is doing. It cannot be used for an ultrasound estimate of weight during pregnancy, because babies who are born preterm are not a fair sample of babies still in the womb at that week. And it describes babies born in 2016 to 2024; average birth weight in the United States has been falling slowly for two decades, so a reference from the 1990s would place the same baby a few percentile points higher.