Housed sheep (n = 6) were offered pairs of turves of 10 cultivars of perennial ryegrass extracted from pure sward plots over three 12-day experiments. Bite rate, bite mass and DM intake from each turf were recorded and preference was determined as the proportion of total DM removed. The ‘force to break leaves’ (FTB) and the metabolisable energy content were also measured and the marginal energy return (MER) of grazing each cultivar was calculated. Cultivars differed in preference (P < 0.01) and FTB (P < 0.05) in each experiment but MER only tended towards significance (P < 0.10) in the first experiment. There was no direct relationship between MER and preference. In one experiment sheep showed a preference for cultivars with lower FTB.