The dose-dependent bone-sparing effects of dietary isoflavones (IF) were investigated in adult (7-month-old) Wistar rats. Forty animals were ovariectomised, allocated into four groups of ten rats each, and immediately treated orally with IF at 0 (OVX), 20 (IF20), 40 (IF40) or 80 (IF80) μg/g body weight per d for 91 d; ten sham-operated (SH) controls received the same diet without added IF. Animals were killed on day 91. Both femoral failure load and total femoral, diaphyseal or metaphyseal bone mineral densities (BMD) were lower in OVX animals than in SH animals. Urinary deoxypyridinoline (DPD) excretion, a marker of bone resorption, and plasma osteocalcin (OC) levels, a marker of osteoblast activity, were higher in OVX animals than in SH animals. Total femoral and diaphyseal BMD and femoral failure load were similar in IF-treated rats and SH rats. Although metaphyseal BMD in IF40 or IF80 rats was similar to that in SH rats, its value was lower in IF20 rats than in controls. The day 91 urinary DPD excretion in IF40 and IF80 rats, but not in IF20 rats, was similar to that in SH rats. Day 91 plasma OC concentrations in IF-treated rats were similar to day 45 values, but were decreased in OVX and SH rats. Thus, daily IF consumption prevented ovariectomy-induced bone loss, both by depressing bone resorption and stimulating osteoblast activity. Moreover, as only the highest IF level induced a weak uterotrophic activity, the optimal IF dose which preserves both cancellous and cortical bone, but exhibits no oestrogen-like effects on the uterus, was 40 μg/g body weight per d.