The reference values of bone mineral density (BMD) were determined in healthy Saudis of both sexes and compared with US / northern European and other reference data. BMD was determined by dual-energy X-ray absorptiometry (DXA) at the lumbar spine and femur including subregions: trochanter, Ward's triangle, and neck, in 1,980 randomly selected Saudis (age range 20-79 years; 915 males and 1,065 females) living in the Jeddah area. Age-related changes in BMD were similar to those described in US / northern European and Lebanese reference data. Decreases in BMD of males were evident (% per year): 0.3-0.8 (lumbar spine), 0.2-0.4 (femoral trochanter), 0.2-1.4 (Ward's triangle), and 0.2-0.7 (femoral neck). Also, decreases in BMD of females were observed (% per year): 0.8-0.9 (lumbar spine), 0.7-0.9 (Ward's triangle), and 0.3-0.7 (femoral neck). Using stepwise multiple regressions that included both body weight and height, the former had 2-4 times greater effect on BMD than the latter. Using the mean BMD of the <35-year-old group the T-score values were calculated for Saudis. The prevalence of osteoporosis in Saudis (50-79 years) at the lumbar spine using the manufacturer's vs Saudi reference data was 38.3-47.7% vs 30.5-49.6 (P<0.000), respectively. Similarly, based on BMD of total femur, the prevalence of osteoporosis using the manufacturer's vs Saudi reference data was 6.3-7.8% vs 1.2-4.7% (P<0.000), respectively. Saudis (> or =50 years) in the lowest quartile of body weight exhibited higher prevalence of osteoporosis (25.6% in females and 15.5% in males) as compared to that of the highest quartiles (0.0% in females and 0.8% in males). The present study underscores the importance of using population-specific reference values for BMD measurements to avoid overdiagnosis and/or underdiagnosis of osteoporosis.