A quality control (QC) program for bone mineral measurements at the proximal femur by dual-energy x-ray absorptiometry (DXA) was designed for the osteoporosis component of the Third National Health and Nutrition Examination Survey (NHANES III). Major elements of the QC program are (1) a QC center for review of all scans, (2) setup procedures for and continuous monitoring of daily QC procedures, (3) reference standards for cross-calibration, (4) longitudinal studies for assessment of instrument stability, (5) monitoring of technologist performance, and (6) training. This report describes the results of the QC program of the first half of this 6 year study, which began in 1988. Measurements were performed on 7376 subjects in three mobile examination centers, which traveled to a new location about every 3 months, a total of 44 locations. A small percentage (3.5%) of all scans were rejected, mostly because of patient motion during scanning, but 33% of the remaining scans required reanalysis at the QC center to refine the location of the regions of interest. Precision in spine and hip phantoms was below 1% at all ROIs. In 535 subjects with duplicate scans, age 20-91 of both sexes, examined in a blinded review, precision for BMD at the femur neck ROI was 3.2% (CV) and 5.1% for Ward's triangle BMD. The central review improved scan quality, increased the number of usable scans, and reduced significantly the range of the mean percentage difference in the duplicate scans. Minor alterations in machine function were observed and corrected. QC results on phantoms were similar to those obtained with stationary instruments in dedicated laboratories. Retrospective analysis of the regression slopes from QC records did not show a need for correction of the data base, indicating that the instruments were stable during the course of the study.