The aim of this study was to evaluate the sprint racing performance of Quarter Horses in Brazil. Estimating genetic parameters, trends and correlations were obtained by single- and two-trait analyses using Bayesian inference (earnings to 2 years of equestrian age, best time and time class at distances of 301 m and 402 m). The data comprised a period of 38 equestrian years (1978 to 2015) with 23,482 sprint race records from 5861 animals. The heritability estimates were of low to moderate magnitude, ranging from 0.10 to 0.37 (single-trait) and from 0.15 to 0.41 (two-traits), and the repeatability was 0.31 to 0.46. The additive, residual and phenotypic correlations between earnings to 2 years of equestrian age and the other traits (best time and time class in distances 301 m and 402 m) were high (−0.95, −0.96, 0.69 and 0.92), low (−0.29, −0.37, 0.26 and 0.27) and moderate (−0.41, −0.47, 0.37 and 0.47), respectively. There is a positive genetic trend for all traits considered. However, evaluation of the last 10 equestrian years (2006 to 2015) showed negative trends (genetic loss) and trends close to zero (genetic stagnation). The design of breeding programs using the earnings trait in conjunction with other racing performance traits can enhance changes in the genetic gains as a whole in speed-racing Quarter Horses. These findings suggest that the traits studied should be included in breeding selection programs for racing Quarter Horses.