Purpose:
The purpose of the study was to evaluate observer repeatability and inter-rater reproducibility of skeletal age assessment using the Tanner–Whitehouse 3 radius, ulna, and short bones (TW3-RUS) method on dual-energy X-ray absorptiometry (DXA) images, and to apply a complementary framework for evaluating observer-dependent measurement quality.

Methods:
Seventy-one elite male youth football players from the Czech Republic were assessed at two time points separated by one year. Skeletal age was estimated from DXA hand scans by an expert and a trained novice rater using the TW3-RUS method. Both observers repeated all assessments after two weeks. Measurement quality was evaluated using the intraclass correlation coefficient ICC(2,1), standard error of measurement (SEM), Bland–Altman analysis with proportional bias testing, bootstrap confidence intervals, and a predefined ±0.5-year decision threshold.

Results:
Observer repeatability and inter-rater reproducibility were excellent at both annual assessments (ICC = 0.965–0.999). The expert maintained consistently low measurement error (SEM < 1 month), whereas the trained novice reduced SEM from 3.7 to 2.5 months after one year of practical experience. Bland–Altman analyses demonstrated only minor systematic differences and no meaningful proportional bias in the overall sample, whereas only a minority of observations exceeded the predefined decision threshold.

Conclusions:
DXA-derived TW3-RUS skeletal age assessment provides excellent repeatability and reproducibility under standardised conditions. Observer experience primarily improves measurement precision rather than ranking consistency. The principal contribution of this study is a complementary validation framework integrating repeatability, reproducibility, precision, agreement, and decision-oriented analyses, providing a methodological basis for evaluating observer-dependent assessment methods in sport science.