A controversy has developed around the question as to whether bone density values from the distal radius can be used to accurately predict risk of future fractures. To address this question, two separate studies were undertaken: (a) Bone density was measured in 460 healthy ambulatory women living in retirement centers in the state of North Carolina; 83% of these women were followed for up to 60 months for occurrence of minimal trauma hip and wrist fractures. Thirty-one minimal trauma fractures were reported in our study population, representing 8% of those followed. The fracture incidence density rate showed a close inverse relationship with incremental changes in bone density at the distal site. Twenty-eight of the 31 fractures were reported in women with bone density values below the 325-mg/cm2 "at risk" value. (b) Bone density values of the distal radius and the lumbar spine from 360 women (aged 18-85 years) from the Chapel Hill area were used to analyze the error in predicting individual spinal density from the distal radial density. Although the overall correlation was high (r = 0.67) and the confidence intervals were narrow, the prediction intervals were quite wide. Thus, prediction of an individual value of spine density from the distal radius density would result in a value with a range too wide to be clinically useful. We conclude that single-photon absorptiometry appears to be a useful tool for screening normal populations of asymptomatic women for prediction of hip or Colles' fracture risk even though it has limited usefulness in prediction of spinal fracture risk or individual values for spinal density.