Urinary urea concentration is an important predictor of NH3 emission from dairy barns. To reduce urinary urea concentration, accurate and precise prediction of urea concentration for different feeding regimes is a prerequisite. The objective of this research, therefore, was to predict urinary urea concentration of a cow using feed characteristics. To compute urinary urea concentration of a cow, we predicted: urine volume; urinary N excretion, using a regression or a mechanistic model; and the relationship between urinary urea concentration and urinary N concentration, which was derived from experimental data. Model results were validated using experimental data. Cows were fed one of nine diets, which was a combination of one of three rumen-degradable protein balances, and one of three roughage compositions. Each diet was repeated once. Measured parameters included herd, diet, and urine characteristics. Observed urinary urea concentration can be predicted with reasonable accuracy from existing models to predict urine volume and urinary N excretion using feed characteristics. The regression model predicted N excretion slightly better than the mechanistic model. In addition, input parameters required for the regression model are recorded at each dairy farm in The Netherlands. This regression model, therefore, can be used by animal nutritionists and producers to determine diets that result in a reduced NH3 emission.