The bone mineral density (BMD) of the lumbar vertebrae L2-4 and femoral neck was determined by dual-energy X-ray absorptiometry (DXA) in 3222 perimenopausal women-a random stratified sample of the population-based Kuopio Osteoporosis Study (OSTPRE). The mean age of the women was 53.4 years (range 47.9-59.6 years). Twenty-nine percent of the women were past users of oral contraceptives (OC) containing 50 micrograms or less of ethinyl estradiol and 7.4% (n = 250) of the women reported OC use for more than 6 years. There was a slight but statistically significant difference between OC users (n = 939) and non-users (n = 2283) in lumbar BMD (1.134 +/- 0.155 g/cm2 v 1.123 +/- 0.161 g/cm2, p = 0.014). A statistically significant difference was recorded also after adjustment for years since menopause, duration of hormonal replacement therapy (HRT) and present weight (p = 0.044). When the analysis was performed among women who had never used oestrogen replacement therapy (n = 1427) and among premenopausal women (n = 387), no differences in BMD were found between OC users and non-users. Similarly, femoral neck BMD did not differ between the groups. This population-based study demonstrated a slightly higher lumbar BMD among past OC users. However, OC users and non-users differed from each other in many behavioral characteristics. Thus, the differences in BMD may be accounted for more by other factors than by past OC use itself. The low-dosage estrogen OCs used today would not be expected to have any positive bone effects among future perimenopausal women.