CDC Pediatric BMI-for-Age Growth Charts: Percentiles 2000 Dataset

1977 NCHS vs 2000 CDC Revision (Kuczmarski et al 2000)

Prior to 2000, pediatric growth reference charts used in the United States were the NCHS (National Center for Health Statistics) 1977 percentile curves, developed by Hamill et al. and published as DHHS Publication No. (PHS) 78-1650. The 1977 NCHS charts were based on four data sources: the Fels Longitudinal Study (Yellow Springs, Ohio, children from 1929 onward), the National Health Examination Survey (NHES) cycles I, II, and III (1960-1970), and supplementary infant weight data from the Ten-State Nutrition Survey of 1968-1970. The 1977 charts did not include a BMI-for-age chart; weight-for-stature and weight-for-age were the only weight status indicators published for pediatric populations.

CDC 2000 Revision Data Consolidation

The 2000 CDC revision was led by Robert J. Kuczmarski, PhD, and a 14-member working group within the CDC's National Center for Health Statistics Division of Health and Nutrition Examination Surveys. The final report was published in May 2000 as Vital Health Statistics, Series 11, Number 246: "CDC Growth Charts: United States" (DHHS Publication No. (PHS) 2000-1696). The consolidated sample after data cleaning and exclusion criteria totaled n=27,940 children and adolescents aged 2-19 years, drawn from five nationally representative cross-sectional surveys: NHES I (1960-62, n=7,316), NHES II (1963-65, n=7,582), NHES III (1966-70, n=8,403), NHANES I (1971-74, n=3,950), and NHANES II (1976-80, n=689 supplement for upper age tail).

Exclusion criteria applied to the source data included: missing weight or stature measurements, biologically implausible values (weight or stature outside ±4 SD of the age-sex mean), missing age or sex data, known congenital or endocrine conditions that affect growth (per parent/guardian report in NHES supplements), and pregnancy status. Racial/ethnic representation in the 2000 dataset was non-Hispanic White 71.4%, non-Hispanic Black 14.5%, and Mexican-American 14.1% — proportions reflecting the US population distribution of the 1960s-1970s survey collection eras. No data from NHANES III (1988-1994) or later cycles was incorporated into the 2000 growth charts; CDC explicitly excluded post-1980 data to prevent the rising secular trend in childhood BMI (documented from the mid-1980s onward) from shifting what were intended as growth references rather than contemporaneous growth targets.

Key Chart Additions and Changes from 1977

Four substantive changes distinguish the 2000 CDC charts from the 1977 NCHS charts: (1) Introduction of BMI-for-age charts as a separate chart set for ages 2-20 years, requested by the American Academy of Pediatrics (AAP) and the Maternal and Child Health Bureau (MCHB) to provide a standardized weight-stature index for older children and adolescents. (2) Extended age range from 0-18 years (1977) to 0-20 years (2000) to cover the 18-20 year adolescent tail not fully represented in the earlier dataset. (3) Separate chart sets for each sex with no age-truncation differences between boys and girls. (4) Shift from visual smoothing of hand-drawn percentile curves (1977 method) to the quantitative LMS smoothing method (Cole 1990) with reproducible statistical output and explicit equivalent degrees of freedom for each fitted curve.

LMS Percentile Calculation Methodology (Cole 1990)

The LMS method (Lambda, Mu, Sigma) was published by Timothy J. Cole in 1990 in the journal Annals of Human Biology (Volume 17, Issue 1, pages 57-65). The method was selected for the CDC 2000 charts following a comparative evaluation of eight curve-fitting approaches, as documented in Appendix B of Vital Health Statistics Series 11 No. 246. LMS was chosen for its ability to simultaneously model centile skewness (L), central tendency (M), and spread (S) as continuous smooth functions of age, and for its closed-form percentile calculation that avoids resampling methods.

The Three LMS Parameters as Age-Dependent Functions

At each integer and half-integer age from 24.0 months through 239.5 months (19 years 11.5 months), three parameters are estimated: Lambda (L, Box-Cox power) controls the skewness of the BMI distribution at that age; when L = 1 the distribution is approximately normal, when L < 1 positive skewness is present, and when L > 1 negative skewness is applied. Mu (M, kg/m²) is the median (50th percentile) BMI value at that age, with separate curves for males and females. Sigma (S) is the generalized coefficient of variation, equivalent to the standard deviation divided by the mean in the transformed space. Each of L(age), M(age), and S(age) is fitted as a cubic smoothing spline with Equivalent Degrees of Freedom (EDF) chosen via penalized maximum likelihood; EDF values for the CDC 2000 BMI-for-age charts were 9 for boys-M, 8 for girls-M, 5 for boys-L, 5 for girls-L, 4 for boys-S, and 5 for girls-S.

Closed-Form Percentile Formula

For any target percentile P, the corresponding BMI value at a given age is computed using the standard normal deviate Z(P) obtained from the inverse CDF of the standard normal distribution: Z(5th) = -1.64485, Z(50th) = 0, Z(85th) = 1.03643, Z(95th) = 1.64485. When L(age) ≠ 0, BMI(P) = M × (1 + L × S × Z(P))^(1/L). When L(age) = 0 (rarely exact, sometimes approached near puberty), BMI(P) = M × exp(S × Z(P)). This formula is exact under the LMS model assumptions and requires no Monte Carlo or bootstrapping. Percentile lookup tables for the CDC 2000 charts were published as SAS and FORTRAN source code by the CDC in 2000, and the identical formulas are used in the [Teen BMI Calculator](../tools/bmi-teen-calculator.html) on VivMetric. For adult BMI calculation methodology, see [BMI Formula Explained](../articles/bmi-formula-explained.html).

Box-Cox Transformation Rationale

The Box-Cox family of power transformations (Box GE, Cox DR, 1964, Journal of the Royal Statistical Society, Series B) provides a parameterized method for converting a skewed continuous distribution to approximate normality. In the LMS context, the power L is estimated simultaneously with M and S for each age slice rather than fixed a priori, allowing the skewness correction strength to vary across the age range naturally. For the CDC 2000 BMI-for-age dataset, L(age) for boys ranges from approximately -1.8 at age 2.0 years to -0.4 at age 12.5 years to -0.8 at age 19.5 years, indicating persistent positive skewness in BMI at all ages, with skewness most pronounced in early childhood and least pronounced at early pubertal onset. For girls, L(age) ranges from approximately -2.1 at age 2.0 years to -0.2 at age 11.0 years to -0.9 at age 19.5 years, showing a similar pattern with greater early childhood skewness than boys.

Smoothed Percentile Curves

The CDC 2000 BMI-for-age chart set (Chart Numbers C-1 for boys and C-2 for girls) includes 10 smoothed percentile curves per sex. The 10 published curves are: 3rd, 5th, 10th, 25th, 50th, 75th, 85th, 90th, 95th, and 97th. The 3rd and 97th curves represent the outer range boundaries used in screening; the 5th, 85th, and 95th curves are the classification boundaries per the BMI-for-age screening categories. The 50th percentile curve corresponds to the age-sex median. The 10th, 25th, 75th, and 90th curves provide intermediate reference bands. Curve spacing is non-equal in BMI units because the underlying distribution is skewed; for example, the vertical BMI distance between the 95th and 97th curves at age 14 is approximately 2.3 kg/m² in boys, whereas the distance between the 50th and 75th curves at the same age is approximately 2.1 kg/m² despite a 25-point percentile difference.

All curves are constrained to be monotonic non-decreasing in the percentile dimension at every fixed age; no percentile crossings occur in the final published dataset. At ages beyond 18 years, the BMI-for-age percentile curves converge toward the fixed adult BMI thresholds: the 85th percentile curve for both sexes approaches approximately 25.0 kg/m² by age 19.5 years, and the 95th percentile curve approaches approximately 30.0 kg/m², which are the CDC adult overweight and obesity classification thresholds, respectively. This convergence was built into the smoothing constraints applied to the upper age tail to provide continuity between pediatric and adult screening references.

Gender Dimorphism Explanation (CDC Report Data)

The BMI-for-age 50th percentile (median) curves for boys and girls diverge at specific ages per the 2000 CDC dataset. During early childhood (ages 2-5 years), the median BMI for both sexes declines from approximately 16.8 kg/m² at age 2 to a nadir of approximately 15.3-15.5 kg/m² at age 4-5 years (the adiposity rebound point), then begins a gradual lifelong increase through the remainder of the age range. Girls exhibit a slightly earlier adiposity rebound onset (median 4.2 years) compared with boys (median 4.8 years) in the CDC 2000 dataset.

Per Table 6 in the CDC 2000 technical report, which documents sex-specific median percentage body fat (%BF) estimates obtained from DXA sub-study correlations with BMI, girls' median %BF at pubertal onset (defined in the report as Tanner Stage 2 breast development, chronological age approximately 10.5-11.0 years for the 50th percentile timing) is 3.2 to 4.1 percentage points higher than boys' median %BF at the equivalent Tanner stage, when computed at the same BMI-for-age percentile rank. The CDC 2000 report explicitly states this is a population-level observation drawn from paired BMI-DXA measurements in NHANES III data and does not alter the BMI-for-age percentile ranking methodology; BMI-for-age is calculated identically for both sexes, and the screening cutoffs (<5th, 85th+, 95th+) apply without sex-specific adjustment beyond the sex-specific percentile curves themselves.

During adolescence, the female median BMI curve rises more steeply than the male curve between ages 11 and 16, with the crossover occurring approximately at age 10.5 years: before age 10.5, boys have a slightly higher median BMI at every age, while after age 10.5, girls have a higher median BMI until approximately age 17.5, at which point the male median BMI exceeds the female median for the remainder of the adult lifespan. This pattern is consistent with the 1977 NCHS data and the 2006 WHO child standards and is preserved across all 10 smoothed percentile curves in the CDC 2000 chart set.

BMI-for-age vs Weight-for-age vs Stature-for-age: Three Chart Types, Use Cases

The CDC 2000 growth chart suite includes three core chart types for each sex, each serving a distinct screening purpose. No single chart type provides complete growth assessment, per the accompanying clinical guidance document (CDC, "Using the CDC Growth Charts," 2002, MMWR Recommendations and Reports, Volume 51, RR-12).

BMI-for-Age (Charts C-1 Boys, C-2 Girls; Ages 2.0-19.9 Years)

Purpose: Primary weight status screening indicator. Computed as weight (kg) divided by stature (m) squared, plotted against chronological age. Use per CDC clinical guide: Classification of underweight, healthy weight, overweight, and obesity screening ranges via the 5th, 85th, and 95th percentile cutoffs. BMI-for-age accounts for both linear growth and body mass simultaneously, making it the preferred chart for weight-related screening in children over 2 years. Limitation per documentation: BMI-for-age is not a direct measure of body fatness; it correlates with DXA-derived percent body fat at r=0.70 to 0.82 in 8-18 year old populations (NHANES III DXA sub-study, n=3,952).

Weight-for-Age (Charts C-3 Boys, C-4 Girls; Ages 0.0-10.0 Years)

Purpose: Underweight screening and general weight tracking. Weight measured in kilograms (or pounds in imperial versions) plotted against chronological age. Use per CDC clinical guide: Identification of failure to thrive and acute weight loss in infants and children through age 10. The 5th percentile serves as the underweight screening cutoff. Limitation: Weight-for-age does not incorporate stature, so it cannot classify overweight or obesity; a child at the 90th weight-for-age percentile may be tall and lean or short and heavy depending on concurrent stature-for-age. Weight-for-age is discontinued at age 10 in the 2000 CDC charts because stature variance increases substantially after pre-puberty, reducing the interpretability of unadjusted weight.

Stature-for-Age (Charts C-5 Boys, C-6 Girls; Ages 0.0-20.0 Years)

Purpose: Linear growth trajectory assessment. Recumbent length (0-23.9 months) or standing stature (2.0-20.0 years) plotted against chronological age. Use per CDC clinical guide: Identification of short stature (<3rd percentile), tall stature (>97th percentile), and abnormal growth velocity via sequential measurements at 6-month or 12-month intervals. Stature-for-age is the most stable of the three charts across time for an individual child, with tracking coefficients (correlation between early and later percentile rank) of r=0.75-0.85 from age 2 to age 18, compared with r=0.45-0.65 for BMI-for-age and r=0.55-0.70 for weight-for-age over the same span.

International Comparison: WHO Child Growth Standards 2006 vs CDC 2000

The World Health Organization published the WHO Child Growth Standards in April 2006 (WHO Multicentre Growth Reference Study Group, Acta Paediatrica Supplement Volume 450, pages 1-101, ISBN 9241591714 for the 2000 WHO Obesity report and ISBN 924159475X for the 2006 MGRS standards). The WHO standards differ from the CDC 2000 charts in both construction methodology and target population.

WHO 2006 Multicentre Growth Reference Study (MGRS) Methodology

The WHO MGRS collected longitudinal growth data from 1,743 healthy children (882 boys, 861 girls) enrolled from birth to 59 months across six geographically and ethnically diverse sites: Davis, California (USA), Pelotas (Brazil), Oslo (Norway), Muscat (Oman), New Delhi (India), and Accra (Ghana). Inclusion criteria required: term singleton birth (37-42 weeks gestation), birth weight 2500-4000 g, no perinatal morbidity, mother non-smoking during pregnancy and lactation, exclusive or predominant breastfeeding for at least 4 months with continued breastfeeding to 12 months per WHO infant feeding guidelines, introduction of complementary foods between 4-6 months, and no socio-economic constraints on growth (defined as access to clean water, primary healthcare, and adequate household income). The 2006 WHO standards are therefore explicitly constructed as normative references describing growth under optimal health and feeding conditions, whereas the CDC 2000 charts are descriptive population references reflecting the full range of US children, including the mix of breastfed and formula-fed infants and the full spectrum of household socio-economic status.

Official CDC Chart Selection Guidance Verbatim

The CDC published the following official statement on chart selection (cdc.gov/growthcharts/cdc_charts.htm, accessed content as reviewed January 10, 2024): "Clinicians should use the WHO growth charts for children aged 0 to <2 years of age. Clinicians should use the CDC growth charts for children and teens aged 2 through 19 years old." The CDC provides a printed clinical wall chart set that combines the WHO 0-23 month charts with the CDC 24-239 month charts (chart numbers 0-1 and 0-2 for combined sex 0-20 stature-for-age, weight-for-age, BMI-for-age, and head-circumference-for-age) for US clinical use.

Quantitative Differences at Common Percentiles (0-59 Months)

At the 50th percentile for stature-for-age at 24 months, the WHO standard is 0.6 cm taller for boys and 0.5 cm taller for girls than the CDC 2000 reference. At the 85th percentile for weight-for-length at 12 months, the WHO standard is 0.4 kg lower (leaner) than the CDC 2000 reference, corresponding to approximately a 3-5 percentage point difference in prevalence estimates of overweight classification between the two systems for 12-month-old children. The CDC 2000 chart set does not include weight-for-length BMI-for-age below 24 months because the 1977 NCHS dataset lacked paired weight-length measurements below 2 years; the WHO 2006 standards fill this gap.

Selected Percentile Data Table (Ages 6, 10, 14, 17)

The following table provides BMI-for-age percentile values (kg/m²) extracted directly from the CDC 2000 BMI-for-age lookup tables at four representative ages: 6 years (pre-school, post-adiposity-rebound), 10 years (pre-pubertal), 14 years (mid-pubertal), and 17 years (late-adolescent, approaching adult thresholds). Percentiles presented: 5th (underweight screening cutoff), 50th (median), 85th (overweight screening cutoff), and 95th (obesity screening cutoff). Values are rounded to one decimal place per the CDC published table format.

At age 17, the 85th percentile for boys (28.2) and girls (29.3) are both below the adult 30.0 obesity threshold, consistent with the design constraint that pediatric percentiles converge gradually toward adult fixed thresholds during the 18-20 age span. The median 50th percentile for 17-year-old girls (23.7) is nearly identical to that for 17-year-old boys (23.6) while the upper-tail percentiles (85th, 95th) are notably higher in girls at this age, reflecting the pubertal dimorphism pattern described earlier. All values in this table are LMS-model-derived point estimates from the CDC 2000 data release and are reproduced exactly as provided in the downloadable SAS datasets available on the NCHS public website.