In this study a predominantly film dosimetric method was used to measure the effective dose from posteroanterior (PA) lumbar spine and proximal femur scans performed on a Lunar DPX-L machine. Because of the very low dose rate in scanning mode, the depth dose data were determined using a stationary detector configuration. The characteristic curve for the film (Kodak TMAT-H) was obtained and depth dose measurements were made using slabs of "solid water". The film was calibrated using a superficial X-ray unit (calibrated against a standard traceable to a national standard). To assess the change in film response with beam hardening at depth, the film was exposed to calibration beams of different half value layer (HVL). The HVL of the DXA beam was determined for surface and depth doses using aluminium filters and a diamond detector (an energy independent device). All measurements were performed three times. Beam size was measured using film, and the scan areas and times were determined by scanning phantoms. The dose from a scan was calculated using Dsc = DTscAb/Asc, where D = dose rate (stationary), Tsc = scan time, Ab = beam area, and Asc = scan area. Organ doses were determined using an anatomical atlas and ICRP 23 female reference. All film measurements had good precision (coefficient of variation < 4%). There was little variation in film sensitivity with change in HVL (< 1% change for the first three HVLs) and consequently no corrections were applied to the depth dose data. Skin entrance dose was 11.5 microGy. Effective dose in females was 0.19 microSv for the PA lumbar spine. For the proximal femur scan, the effective dose was 0.14 microSv (ovaries included) and 0.023 microSv (ovaries excluded) for pre-menopausal and pos-menopausal women, respectively.