Radial bone mineral density (BMD) of 217 white women aged 22-54 years from a single rural community was evaluated in 1984 using single-photon absorptiometry. BMD was measured at a site one-third the distance from the wrist to the elbow, a site that represents predominantly cortical bone tissue. Most of these women (181; 83%) were reexamined 5 years later. The overall average 5 year radial BMD loss was -5.6%. The average rate of loss was -4.5% for women retaining positive estrogen status at follow-up (n = 108) and -7.4% for women who were in negative estrogen status at follow-up (n = 73). Baseline radial BMD measures were highly predictive of the follow-up BMD values (r = 0.80). Women with positive estrogen status and greater baseline BMD also had less BMD change. Greater baseline BMD did not predict the amount of change in women with negative estrogen status. The bone turnover markers osteocalcin and bone-specific alkaline phosphatase were significantly associated with BMD change in women with negative, but not positive estrogen status. There was no conclusive evidence of a "peak age" in the baseline and follow-up BMD measures. Based on rates of BMD change, "peak" bone mineral content appears to occur before age 25 years. Factors significantly associated with lower levels of BMD were menopause without estrogen replacement, nulliparity, smoking, and age at first pregnancy. Factors associated with more bone loss were menopause without estrogen replacement, smoking, shorter duration of oral contraceptive use, and older age. Quetelet index, muscle area, number of lost pregnancies, ever breast-feeding, or calcium intake were not associated with BMD level or its 5 year rate of loss. Physical activity and alcohol intake were not associated with BMD level or change after data were adjusted for age or estrogen status.