Patients attending a routine bone densitometry clinic were scanned on two different densitometers on the same day using a pencil-beam (Hologic QDR1000W) and fan-beam (Hologic QDR4500W) machine. Subjects were scanned at the lumbar spine site and or the proximal femur. The differences in bone mineral density (BMD) between the fan-beam and pencil-beam (QDR4500W-QDR1000W) were determined for all pairs of scans. The mean difference in BMD was also calculated to see if there was a systematic bias between the machines. The mean difference in BMD was -8 mg/cm2 and 25 mg/cm2 at the spine and total hip, respectively. The individual differences in BMD between the two machines were examined to assess if they were significantly greater than measurement error. The percentage of scans classified as significantly different was calculated for the difference in BMD before and after adjusting for the mean difference in BMD. The percentage of individuals classified as significantly different ranged from 17.1-45.0% before adjustment, at the spine and total hip, respectively, and 16.1-22.6% after adjustment. From a clinical perspective, this degree of misclassification is probably unacceptable. These results suggest that scans obtained from a QDR4500W fan-beam system and QDR1000W pencil-beam system should not be compared, with or without adjustment for systematic bias.