Background: Treatment of latent TB infection (LTBI) is essential for preventing TB in North America, but acceptance and completion of this treatment have not been systematically assessed. Methods: We performed a retrospective, randomized two-stage cross-sectional survey of treatment and completion of LTBI at public and private clinics in 19 regions of the United States and Canada in 2002. Results: At 32 clinics that both performed tuberculin skin testing and offered treatment, 123 (17.1%; 95% CI, 14.5%-20.0%) of 720 subjects tested and offered treatment declined. Employees at healthcare facilities were more likely to decline (odds ratio [OR], 4.74; 95% CI, 1.75-12.9; P 5.003), whereas those in contact with a patient with TB were less likely to decline (OR, 0.19; 95% CI, 0.07-0.50; P 5.001). At 68 clinics starting treatment regardless of where skin testing was performed, 1,045 (52.7%; 95% CI, 48.5%-56.8%) of 1,994 people starting treatment failed to complete the recommended course. Risk factors for failure to complete included starting the 9-month isoniazid regimen (OR, 2.08; 95% CI, 1.23-3.57), residence in a congregate setting (nursing home, shelter, or jail; OR, 2.94; 95% CI, 1.58-5.56), injection drug use (OR, 2.13; 95% CI, 1.04-4.35), age ≥ 15 years (OR, 1.49; 95% CI, 1.14-1.94), and employment at a health-care facility (1.37; 95% CI, 1.00-1.85). Conclusions: Fewer than half of the people starting treatment of LTBI completed therapy. Shorter regimens and interventions targeting residents of congregate settings, injection drug users, and employees of health-care facilities are needed to increase completion.