ABSTRACT The aim of this study was to determine the effects of pretreatment with reducing sugars or with an enzymatic cocktail before extrusion of fava bean on intake, milk yield and composition, N partitioning, and plasma and ruminal parameters. The main hypothesis was that these pretreatment conditions would allow better N protection in the rumen compared with classic pretreatment before extrusion, thanks to an increase of sugar proportion, either exogenous or endogenous with enzymatic actions. Sixteen Holstein cows were used in a 4 × 4 Latin square design experiment. Cow were fed a diet with a 56:44 forage to concentrate ratio and containing 16.2% of crude protein (CP; dry matter basis). Concentrate consisted of fava bean:linseed blends (90:10%, raw basis) distributed either raw, extruded after pretreatment without additive, extruded after pretreatment with reducing sugars, or extruded after pretreatment with an enzymatic cocktail. The experimental blends provided 53% of total CP in the diet. Intake, milk yield and composition, ruminal pH, volatile fatty acids and ammonia kinetics, apparent total-tract nutrient digestibilities and N partitioning, Maillard compounds in feed and feces, plasma AA, and 15N natural enrichment were measured. Data were analyzed using analysis of variance according to the MIXED procedure of SAS (SAS Institute Inc., Cary, NC). Extrusion without additive during pretreatment led to higher contents of Maillard compounds in the blend, a lower enzymatic CP degradability, a numerically lower ammonia content in ruminal fluid, and a trend of higher plasma EAA concentration, all suggesting a decrease in degradability of proteins in the rumen, and a subsequent increase in metabolizable protein supply. Compared with pretreatment without additive, adding reducing sugars or an enzymatic cocktail during pretreatment led to an increase in Maillard compound contents in the extruded blends, and to an increase (+50 mg/L) in rumen ammonia content. With reducing sugars, digestibility of Nɛ-carboxymethyl-lysine was numerically lower, and plasma EAA concentration tended to decrease, suggesting an overprotection of proteins in the intestine. With the enzymatic cocktail, plasma EAA concentration was similar than without additive during pretreatment, suggesting a similar metabolizable protein supply with both treatments. Finally, no change in N partitioning between milk, urine, and feces was observed whatever the diet.