Pediatric dual-energy X-ray absorptiometry spine scans often cannot be analyzed with standard software due to a failure to identify the bone edges of low density vertebrae. Low density spine (LDS) software improves bone detection compared with standard software. The objective of this study was to compare bone mineral density (BMD) measurements obtained with the standard and LDS software in 27 healthy nonobese, 32 obese, and 41 chronically ill children, ages 2-18 years. Lumbar spine (L1-L4) BMD, measured by standard analysis, ranged from 0.531-1.244 gm/cm2. Reanalysis with the LDS software resulted in a systematic increase (mean +/- SD) in estimated bone area of 17.0+/-5.0%, an increase in bone mineral content of 6.1+/-6.3%, and a mean decrease in BMD of 8.7+/-1.7% (all p < 0.001). This resulted in a mean decrease in BMD Z score of 0.7+/-0.2. Linear regression models, predicting standard BMD from LDS BMD, were fit for the three subject groups (R2 = 0.993-0.995). Small differences in slopes were detected across groups (p = 0.07); LDS BMD predicted higher standard BMD in obese subjects. In conclusion, LDS analysis resulted in a clinically significant decrease in measured BMD. The association between analysis methods was exceptionally high (R2 > 0.99), indicating that LDS BMD accurately predicts standard BMD. Although LDS BMD in obese subjects predicts higher standard BMD results than in nonobese subjects, the small difference is of questionable clinical significance. LDS software is a useful tool for the assessment of BMD in children.