The productivity and stability of cattle production on rangelands depends on the maintenance of a dense and productive perennial grass-dominated resource base, which is contingent on appropriate grazing and recovery periods. We investigated the effect of simulated trampling, dung inputs, frequency of defoliation in the previous growing season (grazing history), and timing of recovery periods on various grassland functional responses in two experiments in western and northwestern Botswana. A field-based clipping experiment at the individual tuft scale demonstrated that perennial grasses are most productive when rested for a full growing season, but that productivity of the highly palatable soft leaved Brachiaria nigropedata Ficalho & Hiern. decreases exponentially with increasing clipping frequency in the previous season (a lagged effect of grazing history). This species was also more productive in the next season when rested during the early than late growing season. The less palatable needle-leaved Stipagrostis uniplumis Licht. ex Roem. & Schult. was less resistant to defoliation than B. nigropedata and decreased equally at each clipping frequency regardless of season. A second field-based experiment at the plot scale demonstrated that a full-season recovery period increased tuft densities while its combination with dung increased cover. The effects of hoof trampling on sandy nutrient-poor grasslands appear to be less significant compared with grasslands on fertile soils. Thus, optimal livestock management strategies should aim to promote season-long grazing of both palatable and unpalatable species to disadvantage the less grazing-tolerant unpalatable species and full growing season recovery periods to ensure optimal recovery and future productivity.