It is known that among women over the age of 65, bone mineral density is lower, and the risk of hip fracture higher, in smokers than non-smokers. We report a study in 1334 health pre- and post-menopausal women aged 35-64 years, to determine whether this effect can be attributed to lower oestrogen levels in smokers. Among 676 premenopausal women forearm bone density was no lower in smokers (95% confidence interval 1% lower, 4% higher). Among 543 postmenopausal women who had not used hormone replacement therapy (HRT) for more than a year there was no statistically significant difference, but the lower confidence interval was consistent with a lower bone density in older smokers (by 8% at age 55-59, 16% at age 60-64). Measurements in 194 postmenopausal women not taking HRT showed that oestrone and oestradiol were similar in smokers and non-smokers, as were cortisol and FSH, LH and prolactin. Meta-analysis of the present study and previous studies confirmed significantly higher levels in smokers of the androgens DHEAS (by 37%) and androstenedione (by 34%). Oestrogens were no lower in smokers, and the lower confidence limit excluded more than a trivial effect of smoking in lowering oestrogen. These results indicate that the recognised lower bone density in elderly smokers cannot be explained by an effect of smoking on oestrogen, since in premenopausal women bone density is no lower in smokers and in postmenopausal women oestrogens are no lower in smokers. The data suggest a balance between higher androgen levels but lower rates of conversion of androgens to oestrogens in smokers. The effect of smoking on bone may be due to impaired response of bone and other target organs to oestrogen, or to actions independent of oestrogen.