To evaluate portable bone ultrasound as a technique for the assessment of osteoporosis in disabled and elderly subjects we have used hemiparesis following stroke as a model for disuse osteoporosis. We used the CUBA clinical system in a cross-sectional study of 38 stroke unit and day hospital patients, at an average of 8 months after stroke. On average broadband ultrasound attenuation (BUA) was lower by 5.5 dB MHz(-1) in the hemiparetic (mean 65.9 dB MHz(-1)) compared to the normal limb (mean 71.4 dB MHz(-1)). This was significant at P<0.05 on paired t-test, but appears to reflect the presence of oedema in the weak leg. A total of 16 (42%) of our subjects had such oedema and these subjects showed a larger BUA difference of 9.2 dB MHz(-1). A total of 22 subjects without leg oedema showed a smaller BUA difference (2.8 dB Mhz(-1)). This result is similar to the difference in bone density described in previous work using dual-energy X-ray absorptiometry (DXA) after stroke, but did not achieve statistical significance (P=0.4). Bone ultrasound scanning was simple and acceptable to our subjects, but sensitivity to the presence of oedema may limit its precision in work with disabled and elderly subjects.