Abstract. Current interest in small-scale species dynamics has led to a proliferation of mobility indices. We advocate the use of direct measures of mobility such as immigration rate, extinction rate, residence time, and carousel time. We also demonstrate that the null expectation of cumulative frequency under different null models can be calculated explicitly. Species can depart from the commonly-used 'random reassignment' model simply because of longevity, and not mobility per se. We therefore prefer a random immigration null model, which assumes that immigration locations are randomized.We examined mobility patterns of selected plant species, studied in 256 quadrats of each of four grains (ranging from 1/64 m 2 to 1 m 2 ) in an Oklahoma grassland. Residence times and carousel times can be centuries or even millennia for some species. We explore the numerical and biological reasons for relationships between mobility statistics. Mobility statistics are fairly consistent among grains and years, although the residence times of species exhibit some subtle scale dependence. Species depart from a random immigration model very slightly -but the departure is consistent: species tend to re-occupy previously vacated space more often than expected due to chance. We believe that the use of direct indices will facilitate the study of how species characteristics influence mobility.