Dual X-ray absorptiometry (DXA) has currently become a clinical standard for the assessment of bone mass and bone mineral density (BMD) at multiple sites for the diagnosis and follow-up assessment of osteoporosis in humans. The precision of DXA measurement in human studies has been well documented during the last two decades. However, there have been no systematic reports on the precision and accuracy of BMD measurements in mice using DXA, although mice have proven to be useful models for the study of osteoporosis. Accordingly, BMD of total body as well as regions of interest (ROIs) was measured twice in mice in vivo after a short (10-min) and long (16-hr) interval between scans by DXA, and scanning variations were calculated. Inter- and intra-analyzer variations from the same scans were also determined. The percent coefficients (%CVs) of short-interval scanning variation and inter- and intra-analyzer variations for total body and regional BMDs were less than 2% at sites, demonstrating high precision of in vivo BMD measurements in mice. Moreover, the BMD values comparing in vivo and ex vivo samples from the same animals were of %CV less than 10% at all sites. The correlation of bone mineral content (BMC) to bone ash was further examined, and the correlation between ROI BMC and bone ash was relatively high at all sites both in vivo and ex vivo, with the latter higher. We conclude that in vivo DXA BMD measurements in mice are very reliable with high precision and acceptable accuracy, and therefore useful for longitudinal studies of the mouse skeleton.