Purpose:To test the concurrent validity of data from 2 different global positioning system (GPS) units for obtaining mechanical properties during sprint acceleration using a field method recently validated by Samozino et al.Methods:Thirty-two athletes performed maximal straight-line sprints, and their running speed was simultaneously measured by GPS units (sampling rate: 20 or 5 Hz) and either a radar or laser device (devices taken as references). Lower-limb mechanical properties of sprint acceleration (theoretical maximal force, theoretical maximal speed, maximal power) were derived from a modeling of the speed–time curves using an exponential function in both measurements. Comparisons of mechanical properties from 20- and 5-Hz GPS units with those from reference devices were performed for 80 and 62 trials, respectively.Results:The percentage bias showed a wide range of overestimation or underestimation for both systems (-7.9% to 9.7% and -5.1% to 2.9% for 20- and 5-Hz GPS), while the ranges of its 90% confidence limits for 20-Hz GPS were markedly smaller than those for 5-Hz GPS. These results were supported by the correlation analyses.Conclusions:Overall, the concurrent validity for all variables derived from 20-Hz GPS measurements was better than that obtained from the 5-Hz GPS units. However, in the current state of GPS devices’ accuracy for speed–time measurements over a maximal sprint acceleration, it is recommended that radar, laser devices, and timing gates remain the reference methods for implementing the computations of Samozino et al.