This paper presents accelerometers as a viable alternative or add-on in the quest for improved athlete assessment techniques. In elite level team sports, global positioning systems (GPSs) are an important tool for workload management. These devices, however, have limitations particularly when it comes to high speeds or indoor environments. Tri-axial accelerometer data were synchronously collected with GPS data from senior elite level Australian Football League athletes. For each athlete (n=44), the accelerometer data were filtered and the step frequency was extracted for periods of constant GPS running speed (>8 km/h). A quadratic fit was observed and applied to each athlete, modeling their step frequency to running speed (mean r^2 =0.8554 ± 0.064). The resulting model showed increased accuracy at higher speeds (speeds above 15 km/h <;10% error and speeds above 20 km/h <;5% error).