1. A semi-arid environment is defined as one in which the effective precipitation is too small for the successful practice of continuous crop production without irrigation water, but sufficient for the seasonal growth of forage and for limited animal production. 2. The direct and indirect results of low rainfall on domestic grazing animals are discussed, and it is suggested that they are additive in their effect and subject the animal to very considerable physiological stress during the dry season. 3. Details are given of experiments, using sets of identical cattle twins managed in the field and the metabolic stall, and planned by the East African Agriculture and Forestry Research Organization to evaluate some of the direct and indirect effects of low rainfall on domestic cattle. 4. In a two year (1960–62) field trial, water deprivation of cattle for 48 and 72 hours, as compared with watering daily, decreased water intake by 11.5 and 17.9%, respectively. Water deprivation for 72 hours significantly depressed liveweight gain, the difference in favour of the daily watered cattle being 14.9%, but the depression in liveweight gain occurred during the first months of the experiment and there was some evidence that the water deprived cattle acclimatised themselve and that they thrived better than could reasonably have been expected. 5. This experiment dramatically demonstrated the importance of the ‘green-grass’ loss in liveweight of ranched cattle managed in the semi-arid areas of East Africa. This loss occurs after the onset of the rains and accounted for 60% of the total loss of liveweight of all experimental groups during the 1961 dry season. 6. Two further experiments are described, in whichBos taurus andBos indicus identical twin cattle managed in metabolic stalls, were subjected to progressively declining planes of nutrition that simulated the decline in the nutritive value of semi-arid grazings during the dry season and finally to water deprivation for 96 hours. 7. In the first experiment it was shown thatBos taurus were able to decrease their urea-N output from 36.1 ± 2.28 to 2.3 ± 0.06 g. daily, and the urea-N proportion of total N- output from 70.2 to 15.0% and from 74.1 to 35.9%, respectively. 8. Data from the second experiment, basically similar in design but planned on a larger scale, showed that total dry matter intake and voluntary water intake were interrelated, and that as the quality of the feed deteriorated, total dry matter intake and voluntary water intake (V.W.I.) declined. At the same time urine and faecal water outputs were reduced, urine water output decreasing first. 9. The experimental data suggested thatBos indicus might possess two characteristics that would give them an advantage overBos taurus cattle in a semi-arid environment. Their dry matter intake did not fall so rapidly and they were more efficient at reducing urine and faecal water output. 10. When the nutrient intake with ‘ab lib’ water was high, the nitrogen balances of all the cattle were very positive, but they fell rapidly as the nutrient content of the ration declined and ultimately became negative. Surprisingly, when 96 hour water deprivation was introduced, the nitrogen balances moved towards the positive end of the scale and in the case of two animals the balances actually became positive. 11. One major implication of these results is that nitrogen balance data is completely dependent on the type of watering regime to which experimental animals are subjected, and that there is no such concept as ‘ad lib’ watering as the animal alters its V.W.I. according to the nutrient content of its feed. Furthermore, for at least one nutrient nitrogen, water deprivation can ensure that an animal is in a state of maintenance when its crude protein intake is far below previously published values. 12. In the final section some suggestions are made as to how the semi-arid grazings of East Africa can best be utilised.