Intervention should be considered in postmenopausal women with bone mineral density (BMD) > or = 1 SD below the reference (T or Z score < -1). However, it is unclear when densitometry should be repeated. This study aimed at determining the need for repeat DXA within 5 years in untreated peri-/postmenopausal women to detect declines of T or Z score to below -1 with 85% confidence. A cohort of 925 healthy women (aged 51.2 +/- 2.9 years) were followed within the Danish Osteoporosis Prevention Study (DOPS) for 5 years without hormone-replacement therapy (HRT). DXA of spine, hip, and forearm was done at 0,1, 2, 3, and 5 years (Hologic QDR-1000/2000). The annual loss in SD units was 0.12 +/- 0.10 at the spine (1.3%), 0.10 +/- 0.09 at the femoral neck (1.2%), and 0.07 +/- 0.09 at the ultradistal (UD) forearm (1.0%). Accordingly, T scores below -1 developed earlier at the spine. The need for a future DXA scan to predict declines of T and Z scores below -1 depended strongly on baseline BMD. In subjects with a positive T score, the risk of developing T < -1 remained at <15% for 5 years at all measured sites. A new scan was needed after 1 year if the T score was below -0.5, and after 3 years if the T score was between 0 and -0.5. Slightly longer intervals apply if Z scores are used. Follow-up densitometry in untreated women should be individually targeted from baseline BMD rather than scheduled at fixed time intervals. An algorithm for planning repeat densitometry in perimenopausal women is provided.