Dimitropoulou et al.[1] reported that frequent sexual activity was associated with increased risk of prostatic cancer in young men (in their 20s) but with decreased risk in older men (in their 50s). This finding largely confirms a conclusion expressed 20 years ago in a review by Bosland [2]: ‘In comparison with controls, prostatic cancer patients (1) have an earlier onset of sexual activity in any sense; (2) show a higher sexual drive especially at a young age and (3) have notwithstanding a lower frequency of intercourse . . . at older ages’.

It is tempting to suggest that this apparent gradual change in behaviour is due to increasing concentrations (with age) of oestradiol-17β (aromatized from androgens) depressing coital rates and acting as a carcinogen [3,4] in men later diagnosed with prostate cancer. It might be possible to test this using data on the sequences of the sexes of offspring of large samples of men diagnosed with prostate cancer. There is good evidence that in mammals (including man), the probability of a male birth is positively associated with paternal testosterone levels around the time of conception [5,6] and, independently, with coital rate [7]. It has been shown that, overall, the offspring sex ratio of men with prostate cancer is not appreciably different from that of controls [8]. I suggest that (if the present hypothesis were correct) then within a large sample of sibships sired by men with prostate cancer (e.g. as in [8], sons should, on average, occur in the earlier birth orders, and daughters in the later birth orders. (The overall tendency for such a phenomenon to occur in very large data sets, of millions, is so small as to be of no immediate concern here, see [9] pp 729–730). To guard against the possibility that sibship size and sex ratio correlate, workers should test sibships individually by using the test of Haldane and Smith [10].