The main objective of this study was to describe longitudinal patterns of spinal bone loss in normal women who undergo a natural menopause. The second objective was to determine if a proportion of women suffer excessively rapid postmenopausal bone loss from the spine. If this was the case it was the aim to devise a means of predicting the woman at excess risk; but if all women lost bone at similar rates, the aim was to document changing loss rates over the first 5-8 postmenopausal years. Responding women in six suburban general practices recalled for cervical smears who had their last menstrual period 9- 36 months previously were invited to participate in a longitudinal study of bone loss and the biochemical markers plasma osteocalcin and urinary hydroxyproline. Sixty-four subjects agreed to participate, a response rate of 80%. In the ensuing 5 years, six received hormone replacement therapy and are not reported on. The main outcome measures were rates of spinal bone loss over 5 years, measured by dual photon absorptiometry, and radial bone loss over the first 2 years measured to quantitative computed tomography. Spinal bone loss was similar between individuals, with 94% of the variability in the data being accounted for by a statistical model that assumed parallel rates of bone loss. A less restrictive model allowing women to have different rates of spinal bone loss accounted for 12% more of the remaining variance in the data than the previous model. However, rates of radial bone loss were more dissimilar between women than rates of spinal loss. The results of the biochemical data collected serially showed that the plasma osteocalcin rose slowly to a plateau at 5 years postmenopause; in contrast, the hydroxyproline fell progressively with time over the whole period of study. These results were interpreted as being consistent with diminishing rates of bone destruction which gradually reequilibrated with bone formation as time passed after menopause.