We examined declines in annual pasture dry matter on offer (DMO) from around peak accumulation in spring through summer-autumn in 14 supplementary feeding experiments across the south coast of Western Australia. Changes in dry matter digestibility (DMD) and concentrations of nitrogen, phosphorus, potassium (K), calcium, magnesium and sulfur were also measured. Plants which germinated in summer-autumn were manually separated, as far as possible, by removing green components from the dry material, and it was assumed all dry pasture had grown during the previous growing season. Measurements were made under grazing (stocking rates varying from 8 to 16.7 young wethers/ha) and in exclosures, and the experiments were carried out at 5 locations over 3 years. Within any experiment, there were no significant effects of supplementary feeding treatments on declines in DMO, DMD or digestible dry matter on offer (DDMO). Highest measured DMo and DMD varied from 2600 to 8600 kg DM/ha and from 63 to 8 I%, minimum measured values ranged from 100 to 1600 kg DM/ha and 45 to 54%, with amounts of DM0 lost varying from 2000 to 8500 kg DM/ha. The declines in DMO, DMD and DDMO under grazing through summer-autumn were described by exponential decay curves. Losses in DM0 and DDMO, as calculated from curve parameters, were positively correlated (P<0.001) with maximum DM0 or DDMO, respectively. Loss and average rate of loss in DMo were also correlated (P<0.01) with total rainfall through the period of measurement. The number of rainy days, stocking rate and botanical composition of pastures in spring were not significantly correlated with these losses under the conditions of these experiments. The percentage rate of decline in DM0 was not strongly correlated with any of these variables and its variation across experiments was small. Change and rate of change in DMD were positively correlated (P<0.001) with maximum measured DMD and negatively correlated (P<0.05) with the percentage of grass in the pasture. It was estimated that most of the lost material was not consumed by sheep. The concentrations of some essential minerals in grazed dry pasture approached, or fell below, their critical concentrations in most experiments. Of the minerals measured, only changes in K concentrations in dry pasture were well described by exponential regressions.