Five subtropical perennial grass species, Cenchrus ciliaris, a Cynodon hybrid, Digitaria eriantha subsp. eriantha, Panicum maximum and Pennisetum clandestinum, were subjected to four levels of water availability in a small plot trial under a rainshelter during the summer growing seasons of 1996/97 and 1997/98. This work was carried out on the Hatfield Experimental Farm of the University of Pretoria, Pretoria, South Africa. The average yields for the tufted species (C. ciliaris, D. eriantha and P. maximum) were lower in the 1997/98 than the 1996/97 season. C. ciliaris, however, produced the highest average yields in both seasons. The average yields of the Cynodon hybrid and P. clandestinum (creeping grasses) reacted differently, in that the average yields were higher in the 1997/98 than 1996/97 season. The dry matter yields of all five grass species at slight water deficits, were not significantly different from yields under conditions of no water shortages, when the soil profile was brought to field capacity at the beginning of the growing season. The implication hereof is, that dry matter yields can be maintained, in conditions where slight water deficits occur, if there is some water stored in the soil profile. C. ciliaris, a notable drought tolerant species, produced yields (11.7-20.0 t ha-1) under non-control conditions (W1, W2 and W3) which were comparable to yields obtained from traditionally irrigated grasses such as the Cynodon hybrid (12.0-15.8 t ha-1) and P. clandestinum (5.6-11.8 t ha-1), under control conditions (W4). P.clandestinum, D. eriantha and P. maximum tended to be better adapted to wetter conditions, while the Cynodon hybrid was also able to produce good yields under both water limiting and non-limiting conditions. Water use efficiency, regardless of the grass species, tended to be better under non-control conditions. Grasses were thus able to produce more dry matter per unit of water under conditions where water became scarce. In vitro dry matter digestibility was not negatively affected by water scarcity, while crude protein content more often than not seemed to be better under control conditions. This could be due to the uptake meganism of nutrients, in which water plays a vital role. The number of stoma per unit area was more under control than water limiting conditions. The expression of the different structures on the leaf surfaces, were not altered by the level of water availability. In the absence of water shortages, the growth rates of C. ciliaris, D. eriantha and P.maximum, were almost twice those of the Cynodon hybrid and P. clandestinum, as measured at the end of two growth cycles in an irrigated field trial. In a pot trial, under glasshouse conditions, with C. ciliaris, the Cynodon hybrid and P.clandestinum, both the level of water availability, and the level of nitrogen were varied. Regardless of species and level of nitrogen, water was still used more efficiently under non-control conditions. Higher levels of nitrogen did, however, improve water use efficiency regardless of level of water available. As with higher levels of available water, higher levels of nitrogen also improved yields. In this trial, the Cynodon hybrid and C. ciliarisalso out-yielded P. clandestinum.<,/p> In summary it can be said that C. ciliaris and the Cynodon hybrid are adapted to warm growing conditions in producing high yields regardless of the level of water available. D. eriantha, P. maximum and P. clandestinum would produce better in such growing conditions if water was not as limiting. Regardless of the species, the plants were able to use water more efficiently under non-control conditions, while in Vitro digestibility was not negatively affected by water limiting conditions. Due to the uptake mechanism of nutrients, water limiting conditions may, however, decrease the crude protein content of the grass plants. Higher levels of nitrogen had a positive impact on dry matter yields and water use efficiency.