Density of the radius (by single-photon absorptiometry) and the spine (by dual-photon absorptiometry) was measured in 281 normal young women and in a large heterogeneous group of 1,622 consecutive female patients. Both spinal and radial density were useful measurements when performed on patients with medical indications; spine and radius densities in these patients averaged 10-20% below those of age-matched controls. However, patients younger than 60 years of age had deficits of spine density that were twice as large as those in the radius, and twice as many of these patients had spinal osteopenia, compared with those with radial osteopenia. After age 70, loss of density in the radius approximated that in the spine and the prevalence of osteopenia was similar at both sites. The poorer sensitivity of the radius measurement reflected the fact that it did not predict spine density. The standard error of the estimate was 0.12 gm/cm2 in normal subjects and 0.15 gm/cm2 in patients. At any given radial density, the spinal density of patients averaged 0.18 gm/cm2 (15-20%) below that in normal subjects. Predictions of spinal density from body weight were as accurate as predictions from radial density. Preferential spinal osteopenia occurring without radial osteopenia precluded radial densitometry as a screening method for individuals with osteoporosis before age 60. However, measurement of the radius was more effective than body weight for defining groups at risk, even though the radius did not reliably predict individual spine density. A radius shaft density less than 0.55 gm/cm2 occurring with a body weight less than 55 kg was associated with a high prevalence of spinal osteopenia.