This study examined how sheep develop a learned aversion to the annual legume biserrula (Biserrula pelecinus L.). Merino sheep were grazed on paddocks sown with either (i) biserrula, (ii) biserrula+ryegrass (Lolium rigidum Gaud.), (iii) crimson clover (Trifolium incarnatum L.), (iv) crimson clover+ryegrass or (v) ryegrass, for 4weeks. Relative preference (Chesson-Manly selection index) was determined by offering randomly ordered monoculture plots of the three forage species. The groups of sheep were then combined into a single group and grazed a paddock containing a monoculture sward of each plant type for 5d. Relative preference of sheep was tested again after the completion of the combined grazing. The experiment was repeated at three stages of plant phenology: vegetative, reproductive and senesced. Relative preference for biserrula was lower in sheep that had grazed the paddock sown with only biserrula compared with sheep that grazed a combination of biserrula+ryegrass, at both the vegetative (; -011 vs. 032) and reproductive (; 018 vs. 063) stages. At each stage of phenology, when groups were combined and grazed together, preference for biserrula became uniform. This supported the hypothesis that sheep develop an aversion to biserrula when they graze a pasture containing biserrula. However, we concluded that the familiarity of sheep with biserrula did not result in low relative preference; rather, the aversion was more likely a response to the high proportion of this plant in their diet. Social facilitation may have enabled sheep to overcome rapidly the food aversion.