[Truncated abstract] With the rapid development of multiple herbicide resistant weeds in crops, it is likely that an important role now exists for new grazing management strategies in farming systems to provide an integrated approach to weed management. In this thesis we examined the general hypothesis that sowing a legume of low preference by sheep relative to the target weeds of crops would improve the control of those weeds in a grazed pasture. To test this general hypothesis, legumes of low preference by Merino sheep were identified and a series of experiments conducted to determine the effect on pasture composition when these less preferred legumes were incorporated into a grazed pasture. We found a learned response that altered forage preference by sheep was important in determining the effectiveness of grazing to reduce seed set by weeds of crops. Investigations on this aspect of the grazing behaviour of sheep were a key part of this thesis. The short-term relative preference of Merino hoggets among 15 pasture legumes, 4 grain legumes and annual ryegrass was determined by offering adjacent monocultures of each of the forage genotypes to the sheep. The relative preference of the hoggets for each of the 20 forages was determined at three phases of plant growth from estimates of the amount of forage consumed. Sheep showed a low selective preference for Vetch (Vicia sativa L.), chickpea (Cicer arietinum L.), biserrula (Biserrula pelecinus L.), lotus (Lotus ornithopodioides L.) and snail medic (Medicago scutellata L.) cvs. Kelson and Sava at the vegetative phase of plant growth. An indoor method was also developed to test the relative preference of sheep among forages growing in pots. Using this method chickpea and snail medic, but not biserrula, were found to have a low relative preference by sheep at the vegetative phase.