Mature crossbred wethers, grazed for 11 weeks on oat stubble at a stocking rate of 20 sheep ha-1, were unsupplemented or offered supplements of a urea-mineral block, a molasses lick or a urea-molasses lick. Following rain in the fourth week, half of the area was sprayed with desiccant herbicide to establish weed-free stubbles. The stubble initially yielded c. 2600 kg DM ha-1 of dead crop residues plus 900 kg DM ha-1 of fallen grain; after 11 weeks this had fallen to c. 1400 kg DM ha-1 of dead crop residues. Initially the sheep selected diets of 0.03-0.22 grain and 0.78-0.97 dead herbage and, during this time, urea-supplemented sheep selected diets lower in grain and nitrogen (N) content than those selected by unsupplemented sheep. On weedy stubbles, where green herbage yielded more than 100 kg DM ha-1, sheep selected a diet containing c. 0.78 green herbage (N content 2.6%), ate more digestible organic matter (DOM) and gained more weight than those on weed-free stubbles, where the diets consisted of 0.98 dead crop residues (N content 1.1 %). All sheep ate some of the urea-mineral blocks throughout, but 100% acceptance of the licks only occurred by the 6th week. Mean intakes reached desired levels (10 g urea and/or 50 g molasses sheep-1 day-1) thereafter, but individual intakes varied widely on all supplements, with the coefficient of variation generally > 50%. In most periods, supplementation had only small effects on the intake of DOM. During the first 4 weeks, sheep lost weight except those offered the urea-mineral blocks. Over the whole experiment, urea-supplemented sheep lost less weight than unsupplemented sheep, but this was mainly due to higher weight gains on the weedy plots. Mean concentrations of rumen volatile fatty acids (VFA) were < 80 mmol l-1 throughout, and fell to