Stroke can affect the physical, emotional, and social aspects of patients' lives. The purpose of this study was to assess the feasibility and psychometric properties of a telephone-administered version of the Health Utilities Index Mark 2 and 3 (HUI2/3). Subjects included patients who had had an ischemic stroke within the prior 12 months and their unpaid caregivers (n = 76 pairs) and an additional 33 unpaid caregivers of patients who were generally aphasic or severely affected. Complete response rates, test-retest reliability, and convergent, divergent, and known-groups validity were determined. For patient-caregiver pairs, 27% had no complete Health Utilities Index Mark 2 (HUI2) responses (i.e., had missing responses for at least 1 item of each assessment), 51% had partial responses (i.e., had complete responses for at least 1, but not all of the assessments), and 22% had complete responses. For the Health Utilities Mark 3 (HUI3), the percentages were 19%, 52%, and 29%. Test-retest reliability for patients intraclass correlation coefficient (ICC = 0.76 for HUI2; 0.75 for HUI3) and caregivers (ICC = 0.91 and 0.89, respectively) were excellent. There were generally high levels of both convergent and divergent validity. There was limited known-groups validity (mild v moderately and mild v severely affected patients reported different overall HUI2 and HUI3 scores; there was no difference between those with moderate and severe disabilities). The same pattern was found for caregivers. We conclude that the telephone-administered HUI2/3 appears to be reliable and have at least limited validity. However, the proportions of missing data for patient/caregivers administered the HUI2/3 were surprisingly high. This high proportion of missing data would limit the use of the telephone-administered HUI2/3 in the context of stroke trials.