To eliminate unnecessary feeding trials, a mechanistic model of sugarcane digestion was used in the search for suitable supplements to improve milk production. Milk production simulated by the model was compared with data observed in four feeding trials published in the tropical literature where crossbred dairy cows were fed sugarcane/urea diets with different types of supplements. The predicted effects of the supplements on the ruminal microbial population, concentrations of ammonia and volatile fatty acids were also compared with the published results in one experiment. The model indicated the nutrient most limiting milk production for the different feeding situations. The addtion of Leucaena to the basal sugarcane/urea improved the availability of amino acids and long-chain fatty acids, with energy becoming the limiting factor. Supplementation with rice bran increased the availability of energy and long-chain fatty acids, but amino acids then became the limiting factor. Supplementation with both Leucaena and rice bran further improved the milk yield, but availability of energy now limited milk production. Supplementation with Leucaena increased milk production more than supplementation with king grass. The main reason for this increase was increased amino acid absorption due to increased microbial outflow. In all feeding situations, the average difference between the predicted milk production and that observed experimentally was 0.57 kg/d (ranging from 0.08 to 1 kg/d).