The hypothesis that serum concentrations of pituitary hormones, sex steroid hormones, or sex hormone-binding globulin (SHBG) affect the occurrence of prostatic cancer was tested in a consecutive sample of 93 patients with newly diagnosed, untreated cancer and in 98 population controls of similar ages without the disease. Cases did not differ significantly from controls regarding serum levels of luteinising hormone (LH) or follicle stimulating hormone (FSH). Remarkably close agreement was found for mean values of total testosterone (15.8 nmol l-1 in cases and 16.0 in controls), and free testosterone (0.295 and 0.293 nmol l-1, respectively), with corresponding odds ratios for the highest vs lowest tertile of 1.0 (95% confidence interval 0.5-1.9) for testosterone and 1.2 (95% confidence interval 0.6-2.4) for free testosterone. Similar close agreement between cases and controls was found for serum concentrations of estradiol, androstenedione and SHBG, although the mean estradiol level was non-significantly (P = 0.30) lower among cases. Changes secondary to the disease were unlikely to have affected the results materially, since only LH and FSH were associated with stage of disease and this relationship was weak. Our findings suggest that further analyses of serum hormone levels at the time of diagnosis are unlikely to improve our understanding of the etiology of prostatic cancer.