BACKGROUND: Asynchronous video directly observed therapy (VDOT) may reduce tuberculosis (TB) program costs and the burden on patients. We compared VDOT performance across three cities in the United States, each of which have TB incidence rates above the national average.METHODS: Patients aged ≥18 years who are currently receiving directly observed anti-TB treatment were invited to use VDOT for monitoring treatment. Pre- and post-treatment interviews and medical records were used to assess site differences in treatment adherence and patient characteristics and perceptions.RESULTS: Participants were enrolled in New York City, NY (n = 48), San Diego, CA (n = 52) and San Francisco, CA, USA (n = 49). Overall, the mean age was 41 years (range 18-87); 59% were male; most were Asian (45%) or Hispanic/Latino (30%); and 77% were foreign-born. The median fraction of expected doses observed (FEDO) was 88% (IQR 76-96). At follow-up, 97% thought VDOT was "very or somewhat easy to use" and 95% would recommend VDOT to other TB patients. Age, race/ethnicity, annual income, and country of birth differed by city (P < 0.05), but FEDO and VDOT perceptions did not.CONCLUSIONS: TB programs in three large US cities observed a high FEDO using VDOT while minimizing staff time and travel. Similar findings across sites support VDOT adoption by other large, urban TB programs.