The phosphorus 31 nuclear magnetic resonance signal from the wrist bones was studied in controls covering a broad age range and in postmenopausal osteopenic women with or without osteoporosis. The signal, which was obtained in less than 10 minutes, was broad (30 ppm) and easy to distinguish from muscle signals. Results were reproducible in a given subject. Signal intensity increased with the diameter of the wrist. Because of difficulties with in vivo quantification, we were unable to derive an accurate measurement of bone mineral content from the bone signal. Analysis of in vivo relaxation times of bone phosphorus atoms failed to demonstrate significant differences between osteoporotic subjects and same-age controls. These apparently disappointing findings may be ascribable to technical problems, to which solutions may be within reach. Our study provides information on the difficulties of this method and on the modalities of use of 31P nuclear magnetic resonance spectroscopy for evaluation of disorders of phosphorus and calcium metabolism.