To assess the reliability and validity of radiographic absorptiometry of middle phalanges when compared to dual energy x-ray absorptiometry (DEXA) of the spine and hip as an indicator of osteoporosis.DEXA readings from the spine and femur were compared with radiographic absorptiometry measurements obtained the same day in a sample of 50 women (average age 63 yrs; range 23-86).Both spine and femur DEXA scores were significantly correlated with radiographic absorptiometry scores (R = 0.70 and 0.68, respectively; approximate standard errors 0.08). DEXA scores were used as the standard, defining moderate fracture risk by a standardized t score < -2 and marked risk by t < -3. Using cutpoints of t < -2 for DEXA and t < -3 for standardized radiographic absorptiometry values, the sensitivity to osteopenia was 0.62 for DEXA of the spine and 0.63 for DEXA of the femur, with specificities of 0.90 and 0.96, respectively, at these cutpoints; the corresponding false positive and false negative rates for our cohort of women were 0.19 and 0.24 (spine) and 0.07 and 0.24 (femur). Receiver operating characteristic (ROC) curves were plotted, varying DEXA measurement site from spine to femur and standardized cutoff from -2 to -3; the areas under the resulting ROC curves ranged from 0.82 to 0.91.Radiographic absorptiometry holds promise as a practical method for screening for osteoporosis.