Yucca schidigera was fractionated with butan-1-ol, yielding a butanol-extractable (BE) fraction, containing all the in vitro antimicrobial activity, and the aqueous, non-butanol-extractable (NBE) fraction. Four groups of five female rats (12 weeks old) were allowed ad libitum access to diets supplemented with water (control) or 200 mfi kg(-1) total Y schidigera (TOT) or its fraction equivalent of NBE or BE for 64 days. The effects of the fractions and their interactions in the TOT treatment were analysed according to the factorial experimental structure by two-way ANOVA. NBE reduced serum urea (-50%, P = 0.019) and ammonia (-46%, P = 0.037) concentrations, serum/urine concentration quotients of urea (-79%, P = 0.009) and ammonia (-57%, P = 0.002). NBE also reduced hindgut acetate/propionate (-12%, P = 0.007) but increased faecal ammonia concentration (+87%, P=0.039). BE reduced hindgut indoles (-25%, P = 0.023) and interacted synergystically with NBE in the TOT treatment to further reduce hindgut acetate/propionate by 6% (P = 0.006). NBE increased (+27%, P = 0.002) and BE decreased (-57%, P = 0.005) hindgut urease activity levels, resulting in essentially no change (+ 4%) in the TOT treatment. The in vitro antimicrobial activity of Y schidigera is an unlikely explanation for most of its effects in vivo because these are caused by NBE and in vitro antimicrobial activity is exclusive to BE. Sarsasapogenin and smilagenin were also exclusive (> 98%) to BE and cannot account for the effects of Y schidigera on N metabolism. (C) 1998 SCI.