Health preferences have not previously been investigated among patients with tuberculous infection or disease. The authors assessed the reliability of visual analogue scale (VAS) and standard gamble (SG) measurements in this patient population. Participants were interviewed 3 times in English or French, at weekly intervals. They evaluated their own health status, as well as 3 marker health states, derived by consensus among tuberculosis (TB) practitioners. For the VAS, health states were assigned a duration of 6 months, whereas for the SG, the duration was 10 years. One hundred eighty-six potentially eligible individuals were identified from the TB clinic database of the Montreal Chest Institute. One hundred twelve (60%) were successfully located; of these, 106 were confirmed eligible. Sixty-seven (63%) agreed to participate, and 50 completed all study measurements (25 treated for latent TB, 17 treated for active TB, and 8 with previous active TB); 38 out of 50 were foreign-born. Intraclass correlation coefficients for the marker states were 0.57 to 0.81 for the VAS, and 0.75 to 0.87 for the SG. For respondents' own health, the coefficients were 0.60 and 0.87, respectively. Median VAS scores for respondents' own health were 85.0 to 88.0 for the 3 interviews, whereas the median utility score was 97.5 for all 3. Administration of the visual analogue scale and standard gamble instruments appeared reliable in a selected group of tuberculosis patients. Major barriers to recruitment were language and mobility in this largely foreign-born population.