A review of nitrogen balance trials with lactating dairy cows, published in 35 papers (125 different diets, average intake 17.6 kg DM/d, average milk yield 26.1 kg/d) showed excessively high values of N balance (NB = N intake-faecal N-urinary N-milk N), with an average of 38.8 g/d. The distribution of NB had positive skewness (1.19, P < 0.01), with a median value of 24.0 g/d. The possible sources of errors for high NB values reported are discussed, including the volatile N losses from faeces and urine, milk N analysis and the dermal and scurf losses. The corrections, taking these errors into account where necessary, only lowered the average value only to 20.6 g/d (median value of 10.2 g/d). A regression analysis, performed within trial for a smaller data set (14 trials, selected in terms of appreciable differences of digestible N intake between dietary treatments), showed that NB was positively correlated (R2 = 0.444, P < 0.01) with dietary N availability (i.e., digestible N); this result could be interpreted in terms of enhanced overestimation of retained N (or underestimation of mobilisation) as the dietary N availability increases. Improved methodological accuracy in experimental procedures could however reduce NB values, even if other unknown sources of N losses might be accounted.
