Abstract Background Skeletal age (SA) is an estimate of biological maturity status that is commonly used in sport-related medical examinations. This study considered intra-observer reproducibility and inter-observer agreement of SA assessments among male tennis players. Methods SA was assessed with the Fels method in 97 male tennis players with chronological ages (CA) spanning 8.7–16.8 years. Radiographs were evaluated by two independent trained observers. Based on the difference between SA and CA, players were classified as late, average or early maturing; if a player was skeletally mature, he was noted as such as an SA is not assigned. Results The magnitude of intra-individual differences between repeated SA assessments were d = 0.008 year (observer A) and d = 0.001 year (observer B); the respective coefficients of variation were 1.11% and 1.75%. Inter-observer mean differences were negligible (t = 1.252, p = 0.210) and the intra-class correlation coefficient was nearly perfect (ICC = 0.995). Concordance of classifications of players by maturity status between observers was 90%. Conclusion Fels SA assessments were highly reproducible and showed an acceptable level of inter-observer agreement between trained examiners. Classifications of players by skeletal maturity status based on assessments of the two observers were highly concordant, though not 100%. The results highlight the importance of experienced observers in skeletal maturity assessments.