Some possible sources of error when sampling oesophageally fistulated sheep at pasture were examined in eleven trials. Samples collected from sheep asted for varying periods of up to 22 hours did not differ significantly in nitrogen content from samples collected from unfasted sheep. There was no significant difference between the nitrogen content of extrusa collected at the start of a 30-minute period and that of extrusa collected at the end of the period. Sheep transferred from a perennial ryegrass/clover pasture to a native pasture for three months ingested herbage containing 3·80% nitrogen when returned to the original pasture. Control sheep which had remained on the ryegrass/clover pasture selected material containing 3·33% nitrogen. In a further trial sheep transferred from a ryegrass/clover pasture to a Phalaris/ clover pasture consumed herbage with an organic matter digestibility of 70·8 % which contained 3·44 % nitrogen. The material consumed by similar sheep which had grazed the Phalaris/clover pasture for 18 months was 71·5% digestible and contained 3·63% nitrogen.

Diurnal changes in the digestibility and nitrogen content of the diet of free-grazing sheep were observed in two trials. Diurnal changes in nitrogen content were similar on different days but changes in digestibility were relatively smaller and less repeatable between days.

Consistent differences in the composition of the diet selected by individual sheep and selected on different days were observed in a further five trials. Variation between sheep and between days in the nitrogen content of the diet was ±5·6% and ±7·4% respectively, both estimates being expressed as coefficients of variation. Corresponding estimates for digestibility, calculated as standard deviations, were ±1·6 and ±1·3 units. The implications of these results when sampling oesophageally fistulated sheep at pasture are considered.