Abstract
<jats:p>Background. Body fat percentage is considered in sports science as a stable, comparable, and interpretable measure. Athletes are compared to sport-specific reference ranges, values from different instruments are directly compared, and differences between sports are attributed to contact demands. None of these assumptions have been systematically tested, and the reference tables most widely used have not been verified against the primary literature they are thought to be based on. Objectives. To assess whether body fat percentage supports the comparisons typically made with it. Specifically: whether values obtained by different measurement methods are comparable; whether the conventional sport-specific reference ranges originate from primary sources; whether a single percentage has a consistent physiological meaning across the competitive spectrum; and whether contact classification accounts for differences in body composition between sports. Methods. A narrative review was conducted. Sources were identified through PubMed, publisher databases, and references from recent reviews. Each data point was categorized as Tier 1 (traceable to a primary peer-reviewed source with details on sample size, measurement method, and dispersion) or Tier 2 (common practitioner ranges without primary source attribution). Sports were organized using the American Academy of Pediatrics contact taxonomy, which served as the main stratification criterion and was evaluated against alternative classifications. Competitive level was determined using the Participant Classification Framework. Results. Tier 1 data included 5,144 athletes and 228 controls with reported sample sizes, along with 232 data points from a review of 90 studies [1]. Elite male athletes had a mean body fat of 14.1 ± 5.4% (range: 8.7–19.5%), and elite females averaged 21.8 ± 4.1% (17.7–25.9%). Contact level did not align with body type clustering; instead, body mass demand—which is a new classification introduced here—accounted for 68.6% of differences between sports, compared to 17.4% for contact level (permutation p = 0.0017). Measurement method influenced variance more than sex: in basketball, values ranged from 12.4% (skinfold) to 21.4% (DXA), a 9-point difference that exceeds the 7.6-point difference between males and females within the same sport. Combat sports (boxing, taekwondo) clustered at 9.8 ± 2.2%, the leanest group, similar to endurance sports. Collision team sports without weight classes (American football, rugby) averaged 15.9 ± 2.6%. Within collision team sports, positional differences exceeded differences between sports: NFL rosters averaged 17.9 ± 6.9%, with defensive backs around 12% and linemen over 25%. Of 32 ranges examined, none were identified within the search described in Section 2.2 as tracing to primary sources. Decomposition of the gradient at each end of the competitive range revealed two opposing mechanisms: below Tier 2, fat mass index decreased by 29.1% while fat-free mass index was unchanged (−1.2%); between Tier 0–1 and Tier 4–5, lean mass was 52.9% greater while fat mass was 10.8% greater in absolute terms. Coefficient of variation was approximately half as large in sports with low mass demand (20.1%) compared to capped (38.8%) or high (38.7%) mass demand sports. Conclusions. Body fat percentage combines two physiologically opposite processes and thus provides no clear information about which process is dominant: below Tier 2, a decreasing percentage indicates fat loss with stable lean mass, while at Tier 4–5, a lower percentage reflects lean mass gain against increasing fat mass. This offers a mechanistic explanation for why lean mass more reliably indicates training status than adiposity. It also suggests that fat mass index and fat-free mass index should be reported alongside, or instead of, body fat percentage. Contact level is not related to body composition. One possible explanation, proposed here as a hypothesis, is body mass demand—what the sport's mechanics and rules require in total mass—leading to low, high, or capped categories. Methodologically, body composition should not be used as a proxy or covariate for head-impact exposure in neurotrauma risk models, as these are independent factors. Standard sport-specific ranges generally align with Tier 1 data but should be viewed as heuristic rather than definitive values. Two additional analyses (by competitive tier and age) and an applied reference are included. All data are cross-sectional and only show associations. Data for female non-athletes and female sub-elite athletes could not be sourced.</jats:p>