This study investigated biological maturity timing (BMT) in 116 Irish U13 and 958 Swedish U15 male national soccer players using various growth reference datasets. BMT was expressed as a Z-score comparing each player's percentage of predicted adult height (%PAH) to age-specific means and standard deviations reported in the: Berkeley Growth Longitudinal (US), UK 1990 growth reference (UK), and Swedish 2000s growth reference (SWE) studies. Multinomial logistic regression was used to compare BMT classification (Early, On-time, Late) between datasets. One-sampled means t-tests were used to examine BMT-related maturity biases per cohort and reference dataset. For both Swedish U15 (p < 0.001) and Irish U13 (p = 0.003) players, BMT classification was significantly influenced by the selection of the growth reference dataset. Only 61% of players had the same BMT category across all reference datasets. Significant maturity biases existed in favour of early maturing Swedish players (mean = 96.5%PAH) in all datasets, although moderate using SWE (d = 0.7), large using UK (d = 1.02), and very large using US (d = 1.36) (all p < 0.001). Small significant maturity biases existed in favour of early maturing Irish players (mean = 87.1%PAH) using UK (p < 0.001, d = 0.48) and US (p < 0.001, d = 0.33), but not SWE. We conclude that maturity Z-scores must be interpreted with significant caution, given the large variations in BMT classification between reference datasets.