The associations between velocity during a standardized, submaximal 800-m exercise test and blood lactate concentration after exercise were investigated in Thoroughbred racehorses on sand and grass racetracks. Predicted lactate concentrations for each horse's exercise test velocity were calculated from the line of best fit derived from results at each racetrack. The repeatability of the differences between the measured and predicted blood lactate response to exercise was also investigated. Exercise tests were conducted at speeds ranging from 12.8 to 16.6 m/s. The variability of velocity within the exercise test was expressed as the coefficient of variation (CV) of the times for the four 200-m sections.On the sand track, the coefficient of determination for the lactate-speed relationship was highest for an exponential regression equation [lactate (mmol/L) = 0.08e(0.33x)(m/s);r(2)= 0.58, P< 0.05;n = 21]. Similar coefficients of determination were calculated for linear (r(2)= 0.56) and second order polynomial equations (r(2)= 0. 57). On the grass racetrack, the relationship was best described by a second order polynomial equation [lactate (mmol/L) = -0.87x(2)+ 28. 17x - 211.41;r(2)= 0.57, P< 0.05;n = 25]. The mean differences between the measured and predicted blood lactate concentrations in repeated tests were 1.9 +/- 1.8 (SD) and 1.8 +/- 1.7 ( SD) on the sand and grass racetracks. There were no significant associations between the velocity of the exercise and the CV on either racetrack. Differences between measured and predicted blood lactate concentrations, based on track-specific lines of best fit, have potential application in field studies of fitness in Thoroughbred horses.