Puberty stage offers real clues about how much growing a child has left, but it is far less precise than most parents expect. The growth plates in long bones are still active during early and middle puberty, then progressively shut down as estrogen levels climb. Knowing where a child sits in that process helps clinicians estimate remaining growth, yet the timing of key milestones like peak growth velocity varies so much from child to child that puberty stage alone cannot pin down a final adult height with confidence. Predictions become meaningfully more accurate when puberty staging is combined with bone age X-rays, current height, and parental heights.
Why Puberty Stage Matters for Growth
Children grow at a relatively steady pace through middle childhood, then experience a dramatic acceleration during puberty. For girls, the fastest growth typically happens around age 12; for boys, around age 14. This peak in growth velocity is driven largely by rising levels of growth hormone and sex steroids. But here is the catch: puberty stage, as clinicians assess it using a five-point scale of physical development, does not map neatly onto the timing of that peak. A study examining the relationship between pubertal staging and peak height velocity found substantial variability in when the fastest growth occurred relative to a child’s stage of development, which matters when doctors are trying to estimate how much height is still to come.1PubMed Central. Relationship Between Timing of Peak Height Velocity and Pubertal Staging in Boys and Girls Two children at the same visible stage of puberty can be months or even more than a year apart in their actual skeletal maturation.
Data from the Harpenden Growth Study, one of the longest-running longitudinal datasets on child growth, showed that the age at which growth takes off and the age at which it peaks correlated moderately with the timing of physical signs like breast development or genital maturation. But peak growth velocity itself was not correlated with how quickly those physical signs progressed.2PubMed. The adolescent growth spurt of boys and girls of the Harpenden growth study A boy who races through puberty in two years and a boy who takes four years may reach quite different adult heights even if they start at the same stage and the same age.
Estrogen Is the Master Switch
The hormone that ultimately determines when growth stops is estrogen, in both sexes. This surprises many people, since estrogen is commonly thought of as a “female” hormone. In boys, testosterone is converted into estrogen by an enzyme called aromatase, and it is that estrogen, not the testosterone itself, that drives skeletal maturation. Research using aromatase inhibitors in boys with a form of early puberty demonstrated that blocking the conversion of testosterone to estrogen normalized both growth rate and bone maturation, even though testosterone levels stayed in the adult male range.3PubMed. The role of estrogen in bone growth and maturation during childhood and adolescence
Estrogen has a two-phase effect on the growth plates. At low concentrations during early puberty, it actually stimulates growth by boosting growth hormone production. At higher concentrations during late puberty, it accelerates a process called senescence in the growth plate cartilage, wearing out the cells that allow bones to lengthen. Eventually those cells stop dividing entirely, and the growth plate fuses with the surrounding bone.4PubMed Central. Effects of estrogen on growth plate senescence and epiphyseal fusion This is why estrogen’s role is described as biphasic: low doses speed up growth while high doses shut it down.5Journal of Endocrinology. Role of oestrogen in the regulation of bone turnover at the menarche Cases of men born with defective estrogen receptors or an inability to produce estrogen at all have confirmed this: without estrogen’s signal, their growth plates never fused and they continued growing well into adulthood.6PubMed Central. The endocrine role of estrogens on human male skeleton
The practical implication is straightforward. The further along in puberty a child is, the more estrogen has accumulated, and the less growing time remains. But because estrogen levels do not rise at identical rates in all children, two kids at the same puberty stage can have very different amounts of growth plate life left.
How Clinicians Actually Predict Adult Height
There is no single gold-standard method. Instead, clinicians choose among several approaches depending on what information is available and what clinical question they are trying to answer. The most commonly used tools fall into two broad categories: those that require a bone age X-ray and those that do not.
Bone Age Methods
A bone age X-ray of the left hand and wrist lets a doctor compare a child’s skeletal maturity to reference standards. The two most widely used atlases are the Greulich-Pyle method, which involves matching the X-ray to a book of reference images, and the Tanner-Whitehouse method, which scores individual bones and sums them into a maturity rating. Once bone age is established, predicted adult height can be calculated using tables that account for current height, bone age, and sex.
These methods are useful but far from perfect. A comparison of the Tanner-Whitehouse second and third editions found that roughly a third of boys had prediction errors exceeding four centimeters, and a similar proportion of girls had errors exceeding about three and a half centimeters.7Hormone Research. Is TW3 Height Prediction More Accurate than TW2? Preliminary Data A method using maturity-based growth velocity curves reported that it could predict adult height within about five centimeters for boys and about seven centimeters for girls 95 percent of the time.8The Journal of Pediatrics. Prediction of adult height using maturity-based cumulative height velocity curves Cross-validation of another bone-age-based method found standard errors of estimate ranging from roughly three to five centimeters depending on the child’s age at the time of prediction.9PubMed. Cross-validation of the Beunen-Malina method to predict adult height
So even the best bone age methods carry a margin of error on the order of several centimeters. They get more accurate as a child moves deeper into puberty, simply because there is less remaining growth to be uncertain about.
Methods That Skip the X-Ray
The Khamis-Roche method predicts adult height from a combination of the child’s current height, current weight, and the average of the parents’ heights, using age- and sex-specific coefficients. It requires no bone age determination at all.10PubMed Central. Variation in Methods of Predicting Adult Height in Children with Idiopathic Short Stature The original validation found that its errors were only slightly larger than those of methods that do use bone age.11Pediatrics. Predicting Adult Stature Without Using Skeletal Age: The Khamis-Roche Method For many families, this makes it a practical screening tool, since it avoids radiation exposure and the cost of an X-ray.
The simplest approach of all is mid-parental height, often called the “target height” formula. You average the parents’ heights, then adjust up for boys or down for girls. Recent work has refined this idea, showing that after correcting for parental age and using a multiplicative rather than additive sex correction, mid-parental height can explain about 40 percent of the variance in a child’s final height.12PubMed Central. Accurate Prediction of Children’s Target Height from Their Mid-Parental Height That sounds modest, but it means parental height is the single biggest readily available predictor. The limitation is the wide confidence interval: one analysis put it at roughly plus or minus ten centimeters.13Pediatric Research. Target Height as Predicted by Parental Heights in a Population-Based Study Useful for ballpark estimates, not for clinical decisions about treatment.
When Puberty Timing Is Unusually Early or Late
Height predictions become trickiest, and matter most, when puberty starts well outside the typical age range. In precocious puberty, where sexual development begins before age eight in girls or nine in boys, the early surge of estrogen accelerates bone maturation faster than it accelerates height gain. The child may be tall for their age at first, but the growth plates fuse earlier, potentially leaving less time for overall growth. Treatment with medications that pause puberty is considered appropriate in children whose bone maturation is advancing rapidly and whose predicted adult height is being compromised.14PubMed. Increased final height in precocious puberty after long-term treatment with LHRH agonists: the National Institutes of Health experience
Constitutional delay of growth and puberty sits at the other end of the spectrum. These are otherwise healthy children, most often boys, who are short for their age and start puberty later than their peers. The reassuring clinical narrative has long been that they are “late bloomers” who will catch up. But a study measuring near-final heights in boys with constitutional delay found that they ended up significantly shorter than both their predicted adult height and their genetic target height, falling short by about five to six centimeters on average.15PubMed Central. Evaluation of near final height in boys with constitutional delay in growth and puberty A separate analysis comparing five different prediction methods in children with short stature and constitutional delay confirmed that prediction accuracy varied widely depending on the method used. In boys, some methods underestimated adult height by as much as seven centimeters, while others overestimated it by three centimeters.16The Journal of Pediatrics. Adult height in boys and girls with untreated short stature and constitutional delay of growth and puberty
The lesson here is that standard prediction tools were built on populations of typically developing children. When you apply them to children at the extremes of pubertal timing, accuracy drops and systematic biases creep in. In children with isolated growth hormone deficiency, for example, baseline bone age predictions underestimated adult height by about four centimeters in girls and six centimeters in boys, though accuracy improved over the course of treatment.17PubMed Central. Adult height prediction by bone age determination in children with isolated growth hormone deficiency A newer model developed specifically for children with idiopathic short stature found that conventional methods tended to overestimate adult height in that group, with one method overshooting by nearly four centimeters on average, while the new model slightly underestimated by about half a centimeter.18PubMed Central. A Novel Method for Adult Height Prediction in Children with Idiopathic Short Stature Derived From a German-Dutch Cohort
Population Differences and the Limits of Reference Standards
Most bone age atlases were developed using mid-twentieth-century data from white American or European children. A systematic review and meta-analysis of studies applying the Greulich-Pyle atlas across different ethnic groups concluded that it is imprecise and should be used with caution in Asian male and African female populations.19PubMed Central. Is the Greulich and Pyle atlas applicable to all ethnicities? A systematic review and meta-analysis The concern is not just academic: if a child’s bone age is read as more advanced than it truly is, the predicted adult height will be too low, and vice versa. This can lead to either unnecessary treatment or false reassurance.
Nutrition also shapes pubertal timing and growth outcomes in ways that prediction models struggle to capture. Severe undernutrition delays the onset and progression of puberty, effectively stretching out the pre-pubertal growth period but often at the cost of reduced growth velocity.20PubMed Central. Nutrition and pubertal development A child who is malnourished will have a bone age that looks younger than their calendar age, which can make prediction models suggest a taller final height than the child’s nutritional circumstances will actually allow.
AI-Powered Predictions
Automated bone age assessment has arrived in clinical practice, and it is changing both the speed and the consistency of predictions. The appeal is obvious: human readers of bone age X-rays disagree with each other, sometimes by more than half a year, and that variability flows directly into the height prediction. Automated systems using deep learning eliminate this rater-to-rater inconsistency.
A recent deep learning framework that combined precise detection of bone growth centers with a classification network achieved a bone age error of just 0.25 years compared to expert readings.21PubMed Central. Automated Bone Age Assessment and Adult Height Prediction from Pediatric Hand Radiographs via a Cascaded Deep Learning Framework One commercially available system, BoneXpert, has shown that its predicted heights in children with chronic endocrine conditions align closely with near-adult heights and show better reproducibility than conventional manual readings.22PubMed Central. Near-Adult Heights and Adult Height Predictions Using Automated and Conventional Greulich-Pyle Bone Age Determinations in Children with Chronic Endocrine Diseases
However, AI does not magically solve the fundamental problem. One AI-based system achieved a mean absolute error of about 4.6 centimeters when predicting final adult height.23PubMed Central. Bone Age Estimation and Prediction of Final Adult Height Using Deep Learning A study in girls with central precocious puberty found that the best-performing AI model, combined with Bayley-Pinneau prediction tables, achieved an error margin of about 2.9 centimeters at the end of treatment, which was notably better than other combinations.24PubMed Central. Deep Learning–Based Bone Age Assessment for Predicting Final Adult Height in Girls With Central Precocious Puberty AI removes human reading variability, but the underlying biological uncertainty in translating skeletal maturity into remaining growth still imposes a floor on how accurate any prediction can be.
The Secular Trend in Puberty Timing
Children today enter puberty earlier than their grandparents did, a phenomenon known as the secular trend. A Danish study spanning forty years and more than 157,000 children found that the age at peak height velocity dropped by about half a year over four decades: girls born in the early 1930s hit peak growth around age 12.5, while those born in the late 1960s reached it around age 12. Boys showed a similar shift from about 14.5 to 14.2 years.25PLOS ONE. Forty Years Trends in Timing of Pubertal Growth Spurt in 157,000 Danish School Children Across most European countries, adult height has been increasing by roughly one to two centimeters per decade since the mid-1800s.26PubMed Central. Growth and puberty in German children: is there still a positive secular trend?
This matters for height prediction because many reference standards are based on older cohorts. If today’s children are maturing earlier and growing taller than the populations these tools were built on, systematic errors can creep in. Some researchers have investigated whether endocrine-disrupting chemicals in the environment might be driving earlier puberty, but the evidence is inconsistent. One analysis concluded that the growth pattern seen in most girls with early puberty, a pattern of accelerated growth from infancy onward, is largely constitutional and unrelated to chemical exposure.27PubMed Central. Endocrine-Disrupting Chemicals and Early Puberty in Girls Whatever the cause, earlier puberty means earlier epiphyseal fusion, and the net effect on adult height depends on whether the growth spurt compensates enough for the shorter growing period. Generally it does in well-nourished populations, but early puberty is still associated with somewhat shorter adult stature compared to later maturers.28Journal of Endocrinology. Male pubertal development: are endocrine-disrupting compounds shifting the norms?
Height Prediction in Youth Sports
Outside the clinic, one of the most active applications of puberty-based height prediction is in youth athletics. Coaches and sports organizations have begun using a concept called bio-banding, which groups young athletes by biological maturity rather than calendar age. The maturity estimate of choice is the percentage of predicted adult height a child has already reached at the time of observation.29PubMed. Bio-Banding in Youth Sports: Background, Concept, and Application
The reason is practical: in any group of 13-year-olds, some will be biologically close to adult size while others are still pre-pubertal. If you select athletes based on who performs best right now, you end up favoring early maturers, who are temporarily bigger and stronger, over late maturers who may have equal or greater long-term potential. Bio-banding attempts to level that playing field. Players and coaches who have participated in bio-banded competitions have generally reported positive experiences, though the approach reduces maturity-associated variation without eliminating it entirely. The prediction models used in this context are the same ones used clinically, with the same margin of error, which means a late-maturing child’s predicted adult height could still be off by several centimeters. Still, even an imperfect maturity estimate beats grouping exclusively by birth year.
What Height Prediction Cannot Tell You
A common misconception is that a height prediction is a diagnosis or a destiny. It is neither. It is a statistical estimate with a confidence interval wide enough to drive meaningful uncertainty. When a doctor tells a family that a child’s predicted adult height is 170 centimeters, what they mean is that the child’s final height will likely fall somewhere in a range around that number, often plus or minus five centimeters or more.
Families sometimes worry that falling short of a predicted height will cause lasting psychological harm, particularly when growth hormone treatment is involved. A long-term study following both treated and untreated young people with short stature found that treatment did not produce lasting positive or negative psychosocial effects, and that both groups adjusted well into adulthood regardless of their final height.30PubMed. Long-term psychosocial consequences of hormone treatment for short stature This is worth keeping in mind when the conversation around height prediction starts to feel high-stakes: for the vast majority of children, the emotional trajectory through adolescence depends far more on factors other than the number on the tape measure.
The tools keep getting better. AI is removing human reading error from bone age X-rays, new prediction models are being calibrated for specific clinical populations, and the understanding of how estrogen, growth hormone, and genetics interact to set final height is more detailed than ever. But biology is messy. A child’s growth plate does not consult a table before it fuses. Prediction is, and will remain, an informed estimate rather than a guarantee.