Responses in ewe and lamb performance to mid pregnancy supplementation and to the method of providing the supplement were examined when Scottish Blackface ewes grazed heather dominant hills in winter. In these experiments, ewes grazed areas with 20% Agrostis/ Festuca and 80% heather at a stocking rate of 2 ewes /ha during mid pregnancy, and were adequately fed in late pregnancy. Treatments were compared where ewes were unsupplemented in mid pregnancy or where a barley based supplement supplying 1.6 - 1.8 MJ ME and 3 - 6 g N /d,mainly as urea, was offered. Mid pregnancy supplementation reduced ewe liveweight and condition score losses, and there was a linear relationship between these two factors and lamb birthweight, which was increased with mid pregnancy supplementation by 10% over three experiments. When areas with 20 - 40% Agrostis /Festuca vegetation were grazed at stocking rates of 1.2 - 2.1 ewes /ha to create a fourfold difference in herbage allowance of Agrostis /Festuca, there was no effect on ewe liveweight change during mid pregnancy, and a response in lamb birthweight was obtained only in 2- year -old ewes. These findings were supported by the small differences found in the concentrations of plasma urea and ruminal ammonia, and in the in vitro digestibility of oesophageal extrusa samples collected from sheep grazing areas of Agrostis /Festuca during the mid pregnancy period. In a further experiment, diet selection differences attributable to herbage mass at the start of the winter were not manifest where Agrostis /Festuca swards had a herbage mass of less than 3,000 kg DM /ha (with approximately 10% green material) and where ewes were grazed at 2 /ha. It was inferred that the benefits of mid pregnancy supplementation could apply to a wide range of heather dominant hills. No gross differences in ewe and lamb performance were found with the use of hand fed pellets, compared with self help feedblocks in mid pregnancy. Voluntary intake of feedblock was related to the composition of the supplement, but was not affected by the siting of feedblocks either on heather or on Agrostis /Festuca areas, by xi hardness or by the use of containers. Considerable variation in daily group intakes of feedblock occurred when voluntary intakes of feedblock were high, and the frequency of replenishment was restricted, but the variation was also high when feedblocks were offered ad libitum. Variability between ewes in feedblock intake was high. There was a significant rank correlation in feedblock intakes by ewes in a group, which suggested that, with an understanding of the factors affecting intake, this variability could be reduced. Social dominance explained 25% of the variation in supplement intake when feedblock was offered ad libitum. Two -year -old ewes were of lower social rank, had lower supplement intakes, and were the most numerous non -feeders. A simple device to identify non-feeders was successfully tested. It was considered that the high coefficients of variation between days and between animals in intake of feedblock might lead to inefficient use of the nutrients supplied.