Forty-eight autumn-calving cows were fed individually during the winter period. Twenty-four cows were offered a complete diet and 24 cows were given the same dietary ingredients as in the complete diet but offered separately. The complete diet was an integral mix of ingredients in the proportion of 500 g silage dry matter (DM) and 500 g compound DM/kg. The compound offered to the cows receiving the separate ingredients was adjusted so that the total compound offered was similar on both feeding treatments. Cows offered the complete diet had a greater intake than those on the separate ingredients, with intakes of 21.4 and 19.1 kg DM/day (s.e. 0.50, P < 0.001) in the first 10 weeks of lactation; and also over the first 18 weeks of lactation, with intakes of 20.1 and 18.6 kg DM/day, respectively (s.e. 0.84, P < 0.01). The extra DM intake was largely silage. Milk yields on both treatments were similar at 31.0 1/day during the winter and there was no significant difference in live-weight gain between the treatments, with values of 22.0 and 19.8 kg on the complete and separate treatments, respectively (s.e. 9.03, NS). Milk fat content was markedly higher with the complete diet at 35.1 g/kg, compared with 31.0 g/kg on the separate treatment (s.e. 0.13, P < 0.001). At current prices, the cost of the extra silage DM on the complete treatment was 10.5 p/day and the improved value of the milk was 15.5 p/day. Another experiment showed no consistent effect on milk fat content when sodium bicarbonate was included in a complete diet. A casual observation on the commercial part of the herd at Bridgets indicated that, when 2 kg of liquid molasses replaced 2 kg of energy from cereal sources in a complete diet, milk protein content in early lactation was raised.