From 6 weeks before calving to the end of lactation 2 groups of 8 cows were fed a ration incorporating a dairy nut containing 29 g/kg urea or an isocaloric isonitrogenous nut containing fish-meal and groundnut. During the 18 weeks after calving, concentrates were fed to appetite. 305-d yields of milk (7023 and 5796 kg) and lactose (345.7 and 273.4 kg) were significantly lower in the urea-fed cows (P < 0·05). Intakes of dry matter, metabolizable energy and digestible crude protein were similar in both groups until 18 weeks after calving when feeding according to yield commenced, but by 6 weeks after calving the urea-fed cows lost only 5% of their immediate post-calving weight compared with 7% in the controls. There was no evidence of ammonia toxicity in the urea-fed cows. Blood glucose concentrations were similar in both groups, but higher serum urea concentrations in the urea-fed cows suggested that there was poor utilization of ingested N. A model proposed by Roy et al. (1977) has been used to calculate the requirements of rumen-degradable protein and undegraded protein (UDP) of the 2 groups from their energy intakes and milk yields. Between 4 and 11 weeks post-calving, dietary requirement for UDP closely matched intake of UDP. It is concluded that UDP intake was a major factor in determining the peak milk yield. Mechanisms by which the degradability of the dietary protein might influence milk yield are discussed.