The current study investigated short-term precision of fan-beam dual X-ray absorptiometry scans using a sample of subjects typically seen in a routine practice. Seventy-two women were scanned twice each in array fan-beam mode at the lumbar spine and 71 at the hip on a Hologic QDR 2000. Sixty-five were scanned at the forearm on a Hologic QDR 4500A in array mode. Each of the two scans was analysed without reference to its companion scan (independent analysis). The second assessment was also reanalyzed using the manufacturer's "Compare" function (compared analysis). Coefficients of variation (CVs) were calculated for the two forms of analysis at all the sites measured. Paired t-tests were performed to see whether the "Compare" function improved precision. The CV at the lumbar spine was 1.1% in both forms of analysis. The "Compare" function did not improve precision significantly at this site. The highest precision at the hip was found at the (Total) site of the hip (1.2-1.0%, independent and compared analyses, respectively) and the lowest at Ward's triangle. The Compare function significantly improved precision at Ward's triangle. The forearm had the best precision of all the sites measured with the Total site of the forearm being highest (0.5-0.7%) and the ultradistal site lowest (1.6-1.8%). The Compare function significantly decreased precision at the mid-distal and Total sites of the forearm. In conclusion, the highest precision was found at the Total site of the hip, the Total site of the forearm, and the spine. The Compare function did not significantly improve precision at the spine or hip (with the exception of Ward's triangle) and decreased precision at the forearm.