The non-selective rotational grazing system has undergone a long and controversial development. The merits of this grazing system, where relatively large numbers of livestock are herded into numerous small paddocks for short timespans with long rests between grazings, have not been formally evaluated in a long-term monitoring experiment in the Karoo. In this study we used exclosures (controls) on a 7 000 ha farm in the Central Lower Karoo, camped into approximately 50-ha paddocks, to evaluate the impact of this grazing system on certain vegetation parameters. We report on the first four years of monitoring, after each of four replicate paddocks had received four treatments (one treatment = 40 - 60 Large Stock Unit grazing days per hectare over a period of 2 - 16 days). Concentrated defoliation with concomitant trampling, dunging, and urinating did not influence the perennial species composition, and cover of this grassy, semi-arid shrubland. Changes over time in plant composition, and cover are explained by annual, and short-term (e.g. quarterly) rainfall rather than by grazing impacts. Ephemerals were not favoured by this grazing system, but litter was more abundant in the treatment than the control areas. The dominant grass,Eragrostis lebmanniana,and shrub,Pentzia incana, are resilient to this defoliation regime, and show signs of compensatory growth. As yet there is no evidence that non-selective grazing increases diversity, but the severe defoliation and trampling may enhance the biomass turnover rate, resulting in more vigorous, and productive plants in the grazed areas. Diversity is unlkely to change rapidly in response to grazing, largely because of the persistence of grazing-tolerant perennials. Concentrating mixed herds of livestock onto small areas with lengthy rests can be a useful tool for 'kick-starting' moribund karoo veld into greater productivity. Several of the impacts hold potential for restoring the rangeland quality of degraded areas, but this needs to be tested.