Four dietary treatments were based on a flat-rate (5 kg/day) of concentrates with ad libitum grass silage. The concentrates were iso-energetic and iso-nitrogenous, based on either barley or unmolassed sugar-beet pulp and either extracted rapeseed meal (RSM) or a 1: 3 mixture of fish and soya-bean meals (F/S). These diets were offered to 61 multiparous Holstein-Friesian cows in a continuous design experiment from lactation weeks 4 to 22. Milk yields tended to he higher with RSM (25·1 v. 23·9 kg/day; s.e.d. = 0·64; P < 0·1), whilst milk fat (38·1 v. 40·0 g/kg; s.e.d. = 0·81) and milk protein (30·4 v. 31·3 g/kg; s.e.d. = 0·41) concentrations were significantly (P < 0·05) lower. There were no significant effects of treatments on the efficiency of conversion of food-nitrogen (N) to milk-N or on N-retention. A lower organic matter apparent digestibility (g/g) was found for RSM-based diets (0·738 v. 0·763; s.e.d. = 0·0096; P < 0·05). The diets were also offered to four fistulated dairy cows in a Latin-square-design experiment. Concentrate energy source had significant effects on rumen pH (P < 0·05) and ammonia-N concentration (P < 0·01), whilst protein sources had no effect; values were always in the optimal range (pH > 6 and ammonia-N > 50 mg/l). There was a significant interaction effect (P < 0·05) such that the N-degradability of the whole diet, estimated in vivo, was unaffected by energy source for RSM-based diets but highly dependent on energy source for FIS diets. Microbial protein yield was reduced on the RSM-based diets (179 v. 220 g/day; s.e.d. = 9·6; P < 0·001).