Soybean isoflavones have been proposed to be anticarcinogenic, but their effective doses have not been established. To study their bioavailability, seven women consumed 3.4, 6.9, or 10.3 mumol isoflavones/kg body wt in soymilk in each of three meals of a liquid diet on one of three feeding days that were separated by 2-wk washout periods. Subjects were randomly assigned to doses in a cross-over design. Plasma, urine and fecal isoflavones were measured by reverse phase HPLC. In two subjects, fecal isoflavone recovery was 10-20 times that in the other five subjects. Average 48-h urinary recoveries of ingested daidzein and genistein were 16 +/- 4 and 10 +/- 4%, respectively, at all three doses among the five subjects excreting only small amounts of isoflavones in feces, whereas urinary recoveries of daidzein and genistein in the two subjects who excreted large amounts of fecal isoflavones were 32 +/- 5 and 37 +/- 6%, respectively. Urinary isoflavone excretion was nearly zero in all subjects at 48 h after dosing. Average plasma concentration of genistein at 24 h after the breakfast isoflavone dose in subjects excreting large amounts of fecal isoflavones was significantly greater by 2.5-fold than in subjects who excreted small amounts of fecal isoflavones (P < 0.05). In vitro anaerobic incubation of isoflavones with human feces showed that intestinal half-life of daidzein and genistein may be as little as 7.5 and 3.3 h, respectively. These data suggest that human isoflavone bioavailability depends upon the relative ability of gut microflora to degrade these compounds.