In this study, the cross-calibration of a fan beam DXA system (Hologic QDR-2000) to a pencil beam scanner from the same manufacturer (Hologic QDR-1000/W) is described. The scanners were calibrated by the manufacturer using the same anthropomorphic spine phantom at installation. To verify consistent machine calibration, a group of 69 female subjects, aged 46-75, had anteroposterior (AP) spine and proximal femur scans on the QDR-1000/W followed by pencil and array scans of the same sites on the QDR-2000 during the same visit. Many of the subjects had bilateral examinations of the proximal femur for a total of 123 hip scans. Pencil and array area, bone mineral content (BMC), and bone mineral density (BMD) from the QDR-2000 were compared with the values obtained on the QDR-1000/W, and linear regression equations were derived for relating the two instruments. At the spine, no differences were found between the QDR-1000/W BMD values and the QDR-2000 array BMD values. A slight difference between pencil beam modes was detected but was not deemed clinically significant. Linear regression models relating the QDR-2000 and QDR-1000/W AP spine BMD measurements showed correlation coefficients greater than 0.99, with slopes of 1.00, intercepts equivalent to zero, and small root mean square errors. Comparisons at the proximal femur showed equivalency at the femoral neck and trochanter regions for the two machines in pencil mode, but slight increases in BMC and BMD at the other femoral sites on the QDR-2000 in both pencil and array mode. Correlation coefficients were 0.97-0.99 for all measurement regions except for Ward's.(ABSTRACT TRUNCATED AT 250 WORDS)