We examined changes in estimated metabolisable energy (ME), crude protein (CP) and neutral detergent fibre (NDF) concentrations of irrigated annual pastures through winter and spring in 2002. The types of pastures sampled were ‘high’ subterranean clover (HS, at least 800 g/kg DM clover), ‘low’ subterranean clover (LS, about 400 g/kg DM clover) and ‘low’ Persian clover (LP, about 500 g/kg DM clover). Estimates were also made of selection differentials, namely the nutrient concentration in the pasture fraction likely to be consumed expressed as a proportion of the nutrient concentration in the whole sward to ground level. The ME concentrations in all pasture types increased from May to August, after which concentrations in LS and LP declined, while the ME concentrations in HS declined after September. Low Persian pasture was significantly (P<0.001) higher in ME than the subterranean clover pastures except in September. There was a significant quadratic relationship between ME and time in months and this relationship differed significantly between the 3 pasture treatments. There was a significant (P<0.001) linear decline in CP concentration through the sampling period. The rate of decline in CP concentration was greatest for HS and lowest for LP pastures. Neutral detergent fibre concentrations declined gradually until August and then increased in all pastures. High subterranean clover pastures were lower (P<0.05) than LS pastures in all months. Low Persian clover pastures were higher in NDF than HS pastures, but lower than LS pastures in most months.There were small but significant differences in selection differentials for estimated ME between months when pastures were cut to 4 cm. Selection differentials for ME were between 1.01 and 1.13 across pasture types. Crude protein selection differentials were higher than for ME and varied between 1.11 and 1.46. There were small significant differences in selection differentials between months for NDF (0.74–0.96). Calculated selection differentials at different cut heights indicated that a nutrient gradient existed throughout the annual pasture sward. Pastures cut at high (>8 cm) cut heights had higher ME and CP and lower NDF, than pastures cut to 4 cm. Differences in selection differentials at different cut heights were apparent between pasture types.