Background: Race distance aptitude in Thoroughbred horses is highly heritable and is influenced largely by variation at the myostatin gene (MSTN). Objectives: In addition to MSTN, we hypothesised that other modifying loci contribute to best race distance. Study design: Using 3006 Thoroughbreds, including 835 ‘elite’ horses, which were >3 years old, had race records and were sampled from Europe/ Middle-East, Australia/New Zealand, North America and South Africa, we performed genome-wide association (GWA) tests and separately developed a genomic prediction algorithm to comprehensively catalogue additive genetic variation contributing to best race distance. Methods: 48,896 single‐nucleotide polymorphism (SNP) genotypes were generated from high‐density SNP genotyping arrays. Heritability estimates, tests of GWA and genomic prediction models were derived for the phenotypes: average race distance, best race distance for elite, nonelite and all winning horses. Results: Heritability estimates were high (urn:x-wiley:04251644:media:evj13058:evj13058-math-0001 = 0.51, best race distance – elite; urn:x-wiley:04251644:media:evj13058:evj13058-math-0002 = 0.42, best race distance – nonelite; urn:x-wiley:04251644:media:evj13058:evj13058-math-0003 = 0.40, best race distance – all) and most of the variation was attributed to the MSTN gene. MSTN locus SNPs were the most strongly associated with the trait and included BIEC2‐438999 (ECA18:66913090; P = 4.51 × 10−110, average race distance; P = 2.33 × 10−42, best race distance – elite). The genomic prediction algorithm enabled the inclusion of variation from all SNPs in a model that partitioned horses into short and long cohorts following assignment of MSTN genotype. Additional genes with minor contributions to best race distance were identified. Main limitations: The nongenetic influence of owner/trainer decisions on placement of horses in suitable races could not be controlled. Conclusions: MSTN is the single most important genetic contributor to best race distance in the Thoroughbred. Employment of genetic prediction models will lead to more accurate placing of horses in races that are best suited to their inherited genetic potential for distance aptitude.