Skeletal muscle strength and mass, major contributors to sprint/power athletic performance, are influenced by genetics. However, to date, only a handful of genetic variants have been associated with sprint/power performance. The ACVR1B A allele (rs rs2854464) has previously been associated with increased muscle-strength in non-athletic cohort. However, no follow-up and/or replications studies have since been conducted. Therefore, the aim of the present study was to compare the genotype distribution of ACVR1B rs2854464 between endurance athletes (E), sprint/power (S/P) athletes, mixed athletes (M), and non-athletic control participants in 1672 athletes (endurance athletes, n = 482; sprint/power athletes, n = 578; mixed athletes, n = 498) and 1089 controls (C) of both European Caucasians (Italian, Polish and Russians) and Brazilians. We have also compared the genotype distribution according to the athlete's level of competition (elite vs. sub-elite). DNA extraction and genotyping were performed using various methods. Fisher's exact test (adjusted for multiple comparisons) was used to test whether the genotype distribution of rs2854464 (AA, AG and GG) differs between groups. The A allele was overrepresented in S/P athletes compared with C in the Caucasian sample (adjusted p = 0.048), whereas there were no differences in genotype distribution between E athletes and C, in neither the Brazilian nor the Caucasian samples (adjusted p > 0.05). When comparing all Caucasian athletes regardless of their sporting discipline to C, we found that the A allele was overrepresented in athletes compared to C (adjusted p = 0.024). This association was even more pronounced when only elite-level athletes were considered (adjusted p = 0.00017). In conclusion, in a relatively large cohort of athletes from Europe and South America we have shown that the ACVR1B rs2854464 A allele is associated with sprint/power performance in Caucasians but not in Brazilian athletes. This reinforces the notion that phenotype-genotype associations may be ethnicity-dependent.