Five dairy cows were arranged in a 5 x 5 Latin square design to compare the effects of two amounts of either duodenal glucose or ruminal propionic acid (C3) on milk yield and composition. Treatments consisted of a grass silage-based diet supplemented with glucogenic nutrients either infused in the rumen as a mixture of volatile fatty acids (control) or pure C3 (1.72 and 3.45 Mcal/d) or in the duodenum as glucose (1.72 and 3.45 Mcal/d). Treatments were isoenergetic and isonitrogenous and contained 100 and 115% of energy and protein requirements according to INRA (1989), respectively. Only C3 treatments significantly modified ruminal volatile fatty acid composition and linearly increased C3 percentage (up to 25.5%). Both treatments substantially decreased milk fat yield and content, and linearly increased milk and protein yields. Although no significant differences between glucose and C3 were highlighted for milk yield and composition, it seems that mechanisms involved in milk fat decrease are different. Indeed, whereas C3 treatments decreased fatty acid production in an homogeneous way, short- and long-chain fatty acids decreased and medium-chain fatty acid production increased with glucose treatments. A bibliographical study confirmed that increasing glucogenic precursors (GP) supply curvilinearly increase milk yield, linearly increase milk protein content (+ 0.04% per Mcal of GP) and curvilinearly decrease milk fat content (- 0.14% per Mcal of GP). Thus, it appears important to account for the nature of energy supplied by the ration in formulation.