The Cornell Net Carbohydrate and Protein System (CNCPS) post-rumen digestion model calculates nutrient absorption in the small intestine from microbial cell, feed residue and endogenous secretion composition, but thereafter it only makes use of one class of nutrient, the absorbed amino acids (AA). This is because apparently digested protein, carbohydrate and fat are amalgamated in the calculation of total digested nutrients (TDN) as a measure of energy supply, which is then converted to digested energy (DE), metabolizable energy (ME) and net energy for lactation (NEL). At the maintenance level of feeding, this results in TDN values for feeds 0.5 to 2.5% below those listed in NRC (1988). ME values predicted at a typical production feeding level (L = 4) are about 6% below the listed maintenance level values, when a reduction of only about 5% would be predicted. The model makes no direct use of NRC (1988) tabulated values for the TDN, DE, ME or NEL of feeds. A sensitivity test of the supply model revealed that variations of ±10% in many of the input parameters were without significant effect upon the nutrient supply measured as TDN or metabolizable protein (MP) supply. Variations in dietary crude protein (CP) concentration affected rumen N and peptide supply, and MP supply from undegraded intake protein (UIP). Plasma urea N and milk urea N varied ±20% because of a multiplier effect within the model, twice the imposed CP% variation. Variations in neutral detergent fibre concentration (NDF%) affected effective NDF (eNDF) and microbial protein synthesis. Variations in the rate of degradation of cell walls (NDF) caused a mean change of ±7.7% in microbial efficiency, whereas variations in the degradation rates for the A and B1 carbohydrate (CHO) fractions were without effect on microbial efficiency. Variations in starch content (B1 fraction of CHO) quantitatively affected microbial production, but this was offset by a consequential reduction in the size of the A fraction (sugars and soluble CHO) of CHO, if all other parameters were held constant. Variations in the fat content of the diet produced significant effects upon predicted TDN and ME values, because of the high gross energy of fat and the multiplier (2.25) used on fat percentage in the calculation of TDN values. Published tests of the CNCPS claim that it can predict non-ammonia nitrogen (NAN) outflows from the rumen of dairy cattle adequately. There are weaknesses in the statistical analysis of the NAN data, particularly the pooling of data from growing cattle with that from dairy cows. Later work has shown that NAN supply in the CNCPS is over predicted with high undegraded protein diets, since any shortage of rumen degraded protein does not reduce microbial protein synthesis.