The objective of this experiment was to quantify the effects of unroasted or roasted ground-shelled corn (GSC), when fed with alfalfa ensiled in bag, bunker, or O 2 -limiting tower silos on ruminal digestion and microbial CP synthesis in lactating dairy cows. The roasted corn was heat-treated in a propane-fired roasting system. Alfalfa was harvested as second cutting from fields with regrowth of the same maturity. A portion of each field was allotted to each silo. The diets with 3 3 2 factorial arrangement of treatments were fed to six multiparous rumen-cannulated Holstein cows in a cyclic change-over design with five 21-day periods. Experimental diets were comparable and averaged (on dry matter (DM) basis): 410 g/kg alfalfa silage (AS), 150 g/kg corn silage, 350 g/kg GSC, 50 g/kg soybean meal, 40 g/kg roasted soybeans, 177 g/kg CP, 264 g/kg NDF and 250 g/kg starch. Nutrient flow was quantified by the omasal sampling technique with use of three markers (Co, Yb and indigestible NDF). Continuous infusion of 10% atom excess ( 15 NH 4 ) 2 SO 4 was used to label microbial CP. None of the interactions between storage structure of dietary AS and corn type were significant. DM intake was not different among dietary treatments, averaging 24.5 kg/day across diets. Means of ADF digested in the rumen for cows fed diets with AS from bag, bunker and O 2 -limiting tower silo were 2.1, 1.7 and 2.1 kg/day, respectively, and was lower in cows fed AS from the bunker silo. This response may partly be a reflection of the higher intake of ADF by cows fed AS ensiled in the O 2 -limiting tower silo compared with the bunker. There was a slightly greater supply of fermentable substrates for cows fed diets with roasted compared with unroasted GSC. The small increases in yield of milk protein and lactose observed in the previous production trial in cows fed diets containing roasted corn may have occurred because of greater supply of fermentable substrates.