Multivariate prediction models were developed from the results of three divergent trials with multicannulated steers. Diets differed in composition and physical structure. Energy contents varied from 10 to 13 MJ ME/kg DM, starch from 20 to 62%, CP from 102 to 155 g/kg DM and RDP from 60 to 85%. In the first trial, diets were compiled from byproducts of the milling industry, in the second, maize meal was the major component and in the third high and normal lysine maize cultivars were fed either whole or rolled. Feeding levels were controlled or ad libitum, varying between 65 to 95 g DM/kg W0.75/d. With more starch in the diet, proportionally more passed to and was digested in the lower digestive tract. More starch tended to pass into the duodenum with lower RDP in the diet, but the effect was indirect through a reduction in the proportion OM which was apparently digested in the rumen (OMDR). NAN passage and digestion in the lower digestive tract were negatively associated with starch content of the diet. This effect was apparently rumen pH related, where high starch fermentation in the rumen lowered pH probably to the detriment of microbial protein production. NAN passage to the duodenum was associated positively with N intake or protein content of the diet and negatively with OMDR. RDP level in the diet per se did not affect the amount of NAN passing to the duodenum, but its interaction with OMDR was highly significant. Thus, with lower RDP levels, OMDR was less, resulting in more NAN passing to the duodenum. Multivariate models predicted NAN and starch passage to the lower digestive tract with an r2 between 0.89 and 0.97 and SDs of 15 g NAN/d and 60 g starch/d. Non-intake associated differences between steers in amino acid and glucose availability at the duodenum was substantial, explaining partly the vast difference in feedlot perform¬ance of individual steers.