The quantity and quality changes of pasture in an all grass wintering system on a high country station, Central Otago, were studied at the end of the 1981 winter after an autumn of good pasture growth and during the 1982 autumn and winter after poor growth in the autumn. Higher levels of green herbage survived after the winter in 1981 compared with poor survival in 1982. Less than half of the green herbage present at the beginning of June was utilised during grazing, and losses of green herbage by frosting or senescence during June 1982 ranged from 35 to 70%. Green percentage of herbage was a good predictor of digestibility when values of 100% green and 100% dead material analyses were added to those of mixed herbage samples, with an average of 89% of variation in herbage digestibility being accounted for by the regressions of in vitro digestibility and % green. Prediction of digestibility from % green of mixed herbage samples alone was not as good (r 2 = 0.53 for June and 0.67 for July). Digestibility of 100% green herbage was high (82% mean) and varied little, whereas that of standing dead fluctuated, and was low (40% mean). Little, if any, dead herbage was eaten, and presence of dead herbage was of no advantage in autumn-saved pasture. There were no differences between clovers, grasses, or other species in winter survival or in digestibility of green herbage.