The high inherent variability in rainfall and forage availability in arid regions makes it very difficult for a rancher to establish a herd size that is suited to the environment and also complicates management decision-making regarding the distribution of animals within and between areas of a single ranch. We recorded the temporal and spatial variability in rainfall and grass production on a local scale on 3 ranches in arid Namibia over 3 years, to determine their effects on potential stocking density, stock movement decisions and timing of grazing. We conducted a 3-year experiment to test the effects of the timing of grazing on grass regrowth. Furthermore, we tested the reliability of a rule-of-thumb used by ranchers to ascertain whether a drought year is likely and whether they should sell some of their cattle in advance of drought. We expanded the scale of the study to include 31 ranches over a 700 km-long rainfall gradient to determine the reliability of extrapolating conclusions obtained at a local scale to a larger geographic scale. We found that rainfall and grass production varied widely, resulting in recommended stocking densities up to 10-fold smaller than those currently recommended by the Namibian Ministry of Agriculture. We found that there is no optimal spatial or temporal scale of ranch management, and that ranchers could do no better than a simple rule of “move the cattle to wherever there is most grass”. This simple rule's validity was reinforced by the results of our experiment, which showed that there was no significant effect of timing of grazing on grass regrowth. Statistical analyses of long-term rainfall records indicated that the ranchers' rule-of-thumb regarding approaching drought was unreliable and that this could lead to substantial loss of profit if adhered to.