The results of five experiments are reported as tests of a theory of the growth and feed intake of sheep. An attempt is first made to describe potential growth in the sheep, that is, growth under non-limiting conditions, and to generalise this description across breeds. The two ways of providing non-limiting conditions are then considered. This approach necessarily confronts the problem of predicting the desired intake of feed resources required to fulfill the potential. Whilst ad libitum access to a single,nutrient-rich feed is found an acceptable method, data collected in three experiments, using a total of 64 Suffolk X Greyface lambs, lead to theconclusion that sheep can also attain their potential for growth when given free and continual access to more than one feed, a possible combination of which is non-limiting to growth. It is suggested that the technique may be applied as an independent test of other estimates of resource requirements for growth and to the prediction of feeding behaviour in sheep. The responses of sheep to limiting conditions for growth are studied on the assumption that sheep have rules for partitioning the feed's first limiting resource. The limits to this view are investigated by experiment using 73 Scottish Blackface wethers offered amounts of feed which provide similar crude protein allowances but different yields of energy. It is concluded that the body composition of sheep can be manipulated by nutrition, since the animals will partition scarce energyresources to protein growth before fattening, but that this effect is complicatedin ruminants because protein supply is not independent of feeding level. The theory is then applied to the problem of predicting the recovery of sheep from the effects of growth restraint. An experiment involving 30 Scottish Blackface wethers provides data which support the idea that sheep, following a period of underfeeding, will resume their known potential for growth and, only when necessary, restore their body composition to the state dictated by their inheritance. No compensatory growth is observed. Finally, the work as a whole is discussed in relation to animal science and the sheepmeat industry. Possibilities for future research are also suggested. The experimental data are published in full in the appendices.