This paper reports the results of three experiments designed to attempt to improve the efficiency of milk production from diets based on a 1: 2 dry matter (DM) mixture of grass silage and whole-crop wheat (WCW) harvested at 550 to 600 g DM per kg and treated with 40 g urea per kg DM. In the first experiment a control diet of the forage mixture offered ad libitum with 9 kg fresh weight (FW) per day of a concentrate mix was compared with seven treatments in which the forage or the concentrates were varied. Eight multiparous cows were used in a four-period incomplete change-over design with 4-week periods. Caustic treatment of the WCW increased DM intake (P < 0·001) but tendencies for higher yields of milk and milk protein were not significant. Replacement of 2 kg FW concentrates per day by the same weight of molasses as a 24-h pre-soak of the forage (molasses pre-treatment) or added to the forage at feeding increased DM intake (P < 0·001) but did not increase milk yield or yield of milk solids. Molasses pre-treatment and molasses added to the concentrates reduced milk protein concentration (P < 0·05). Replacement of 2 kg FW concentrates per day by the same weight of ground wheat increased DM intake when added to the concentrates (P < 0·001) but not when added to the forage and milk yield was unaffected by either method of addition. The ground wheat tended to reult in higher milk protein concentrations than the molasses. Increasing the digestible undegradable protein content of the concentrates had no significant effect on food intake or milk production. In the second experiment diet digestibility and energy and nitrogen (N) balance were measured for the control diet and three of the treatments from experiment 1 in four multiparous cows in a 4 ✕ 4 Latin-square experiment with 5-week periods. There were no significant effects on food intake or milk yield. Caustic treatment reduced starch digestibility (P < 0·001) and increased neutral-detergent fibre (NDF) digestibility (P < 0·05) but had no effects on energy utilization. It also reduced N intake (P < 0·01) and urine N losses (P < 0·001) by reducing the ammonia concentration of the WCW. Molasses added to the forage or concentrates reduced milk protein concentration (P < 0·05) and digestibility of both starch and NDF (P < 0·05) but energy utilization was unaffected. In a further Latin-square experiment with the same treatments and four fistulated lactating cows, caustic treatment reduced daily mean ammonia concentration (P < 0·01) and increased daily mean pH (P < 0·01) in the rumen. It is concluded that these treatments did not generally improve the value of WCW although caustic treatment increased milk yield and milk protein yield consistently but not significantly. However practical problems may limit its use on the farm. Urea-treated WCW must be accepted as being a relatively low-energy food although with high intake characteristics, and future work should concentrate on evaluating WCW harvested at an earlier stage of maturity.