Evidence from the literature regarding the sward characteristics which determine ingestive behaviour and herbage intake in the grazing ruminant is contradictory and inconclusive. Variation in sward height has usually been confounded with concomitant changes in sward density, and often digestibility, making objective interpretation difficult. In the current work, a series of grass and cereal swards was produced using d ifferen t seed rates at sowing in an attem pt to obtain a large and independent variation in sward height and density. These swards were grazed within three months of sowing to minimise differences in maturity and digestibility. A few established ryegrass swards were also grazed, and in the second of the two experiments grazing or cutting pre-treatments were employed to increase further the range of sward conditions. Experiment 1, run over two grazing seasons, comprised a series of trials on 33 large plots which were stocked with sheep and cattle (1983) or just sheep (1984). Swards were grazed down over eight days whilst changes in sward canopy structure, ingestive behaviour and herbage intake were measured. The quantity of cattle data collected was limited, but results for the sheep clearly indicated that bite weight had the dominant influence on herbage intake. Bite rate and grazing tim e tended to increase as bite weight and intake declined, both during the defoliation of a sward and when comparing responses between swards. Bite weight was strongly influenced by bite depth, and the sheep generally grazed deeper, taking heavier bites, when the sward was taller and more digestible. The bulk density of the grazed sward stratum had a minor, and unexpectedly negative, e ffe c t on bite weight. Possible explanations are given. A substantial proportion of the variance in both bite depth and bite weight was attributed to undescribed differences between crops. Since indoor feeding trials did not indicate any intrinsic herbage qualities which significantly influenced voluntary intake, these differences probably reflected unmeasured structural variables. Experiment 2, run in 1984, involved a more controlled approach than the large plot trials. Sheep w ere confined in cages and allowed to take only 20 bites from small patches of sward. Measurements of bite weight, depth, area and volume were related to the characteristics of seventeen contrasting swards. Surface height had a strong positive e ffe c t on bite depth, and consequently bite volume and bite weight both increased on taller swards. The variables which determined bite area were less obvious, but within a given grass species bite area appeared to be related positively to surface height and negatively to the population density of grazed plant units. Grazed stratum bulk density, which varied independently from surface height, also had a positive e ffe c t on bite weight. The relative importance of these two key sward variabies in determining bite weight varied with the range of sward heights under consideration. Their effects, however, were independent and additive, producing a pianar joint response surface. The advantages of the new grazing cage technique are discussed and suggestions made for further studies.