Longitudinal studies of osteoporosis in older persons may underestimate bone loss because of a lack of follow-up measurements on subjects too frail to return. The authors addressed this possible bias as part of the population-based Framingham Study; in 1996-1997, they used quantitative ultrasound to assess the bone status of elderly subjects regardless of their ability to return to the clinic. Broadband ultrasound attenuation (BUA) and speed of sound of the calcaneus (heel) were measured in 433 subjects at the Framingham, Massachusetts, clinic and in 167 subjects at their homes or nursing homes. All ultrasound parameters were measured with intramachine coefficients of variation of <6.0%. The mean BUA for those subjects evaluated at the clinic was higher than for those measured at home (9.2% higher for men, p = 0.081; 8.6% higher for women, p = 0.034). After adjustment for age and weight, the differences in BUA were no longer significant. Among the elderly subjects participating in this longitudinal cohort study, those who were unable to return for follow-up were older, weighed less, and had a lower BUA than those who did return, suggesting that longitudinal studies of changes in bone mass with aging may underestimate the true population values.