We hypothesized that lower ovarian and gonadotropin hormone concentrations would be associated with lower levels of peak bone mineral density (BMD) in apparently normally menstruating women who did not exercise intensively and did not report anorexia or bulimia. This hypothesis was evaluated using a case-with-control study design (n = 65) which was nested within a population-based longitudinal study of peak bone mass (Michigan Bone Health Study) with annual assessment in women aged 25-45 years (n = 582). Cases were 31 premenopausal women with BMD of the lumbar spine, femoral neck, and total body less than the 10th percentile of the distribution, where controls were 34 premenopausal women with BMD between the 50th and 75th percentile. BMD was measured by dual-energy X-ray absorptiometry. In addition to their annual measurement, these 65 participants collected first-voided morning urine specimens daily through two consecutive menstrual cycles. The urine from alternating days of this collection was analyzed for estrone-3-glucuronide (E1G), pregnanediol glucuronide (PdG), testosterone, and follicle-stimulating hormone by radioimmunoassay and these values adjusted for daily creatinine excretion levels. Additionally, analyses of daily urine specimens for luteinizing hormone (uLH) was undertaken to better characterize the possible uLH surge. Cases had significantly lower amounts of E1G (p = 0.009) and PdG (p = 0.002) than did controls, whether amounts were characterized by a mean value, the highest value, or the area under the curve, and after statistically controlling for body size. Further, when B-splines were used to fit lines to the E1G and PdG data across the menstrual cycle, the 95% confidence intervals (CIs) about the line for the controls consistently excluded and excluded and exceeded the 95% confidence bands for the cases in the time frame associated with the luteal phase in ovulatory cycles. Likewise, 95% CIs for the LH surge in controls exceeded the fitted line for cases around the time associates with the LH surge. The cases and controls were not different according to dietary intake (energy, protein, calcium), family history of osteoporosis, reproductive characteristics (parity, age at menarche, age of first pregnancy), follicular phase serum hormone levels, calciotropic hormone levels, or by evidence of perimenopause. We conclude that these healthy, menstruating women with BMD at the lowest 10th percentile from a population-based study had significantly lower urinary sex steroid hormone levels during the luteal phase of menstrual cycles as compared with hormone levels in premenopausal women with BMD between the 50th and 75th percentile of the same population-based study, even after considering the role of body size. These data suggest that subclinical decreases in circulating gonadal steroids may impair the attainment and/or maintenance of bone mass in otherwise reproductively normal women.