9704 ambulatory women aged 65 years or older were prospectively studied to determine whether low bone mineral density (osteopenia) was associated with mortality. Bone mineral density was measured at entry to the study by single-photon absorptiometry. 299 women died during a mean of 2.8 years' follow-up. Osteopenia was associated with increased non-trauma mortality, probably because it is a marker for several other adverse factors. Each standard deviation decrease in proximal radius bone mineral density (0.104 g/cm2) was associated with a 1.19-fold increase in mortality (95% confidence interval 1.04-1.36), adjusted for age and duration of follow-up. Diminished bone mineral density at the proximal radius was strongly associated with deaths from stroke (relative risk = 1.74; 95% CI 1.12-2.70), an association that was not confounded by history of previous stroke, hypertension, postmenopausal use of oestrogen, thiazide diuretic treatment, diabetes mellitus, and smoking. Most deaths in women with low bone mineral density are unrelated to the occurrence of fractures-an observation that should be taken into account when estimating the need for and cost-effectiveness of bone-density screening and fracture prevention programmes.