Although global positioning system (GPS) location data have been used to derive animal movement parameters including step length, rarely have these parameters been used to predict animal responses to human interventions. In this study, we tested whether GPS-derived step length of semi-free range cattle is a function of herder presence. The derived step-length model was used to predict herder presence on independent cattle GPS collar data. We also tested whether cattle foraging behaviour is explained by herder activity and vegetation greenness. We used logistic regression to model herder presence as a function of step length and relate cattle behaviour with herder activity and vegetation greenness. The field-based step length model successfully predicted herder presence on GPS collar data. The average predicted frequency of herder presence for the GPS-collared herds was 31%, whilst the field-based GPS frequency was 27%. Herding activities and vegetation greenness also explained different cattle foraging behaviour.