We investigated the effects of consuming an isoflavone aglycone-enriched soya milk containing viable bifidobacteria on urinary isoflavone excretion and percentage recovery. Sixteen postmenopausal women were randomly divided into two groups to consume either fermented or non-fermented soya milk. Each group participated in a double-blind, crossover study with three 14 d supplementation periods, separated by a 14 d washout. Subjects ingested three daily dosages of isoflavone via the soya milk and collected four 24 h pooled urine specimens per supplementation period. Soya milks were prepared with soya protein isolate and soya germ, followed by fermentation with bifidobacteria. Isoflavone levels were quantified using HPLC. Non-fermented soya milks at 20, 40 and 80 mg isoflavone/200 ml contained 10 %, 9 % and 7 % aglycone, respectively, with their fermented counterparts containing 69 %, 57 % and 36 % aglycone (P<0·001). A trend to a greater percentage urinary recovery of daidzein and glycitein was observed among women consuming fermented soya milk at a dosage of 40 mg isoflavone (P=0·13). A distinct linear dose response for the fermented soya milk group (R2=0·9993) compared with the non-fermented group (R2=0·8865) suggested less interindividual variation in isoflavone absorption. However, total urinary isoflavone excretion was similar for both groups (P>0·05), with urinary isoflavone recovery at approximately 31 %. Increasing the isoflavone dosage correlated positively with its urinary excretion, but urinary percentage recovery of isoflavone was inversely related to dosage level. Hence, a modest dosage ranging from 20 to 30 mg/d may provide the most bioavailable source of isoflavone, regardless of whether it is via an aglycone-rich fermented soya milk or a glucoside-rich soya milk.