The present nonrandomized prospective study evaluated whether antimycobacterial therapy for disseminated Mycobacterium avium complex (MAC) could be withdrawn from human immunodeficiency virus-infected subjects who experienced immunologic recovery while receiving highly active antiretroviral therapy (HAART). Eligible subjects had received macrolide-based therapy for least 12 months, were asymptomatic for MAC, had received HAART for at least 16 weeks, and had CD4+ T cell counts >100 cells/microL. Forty-eight subjects were enrolled, with a median CD4+ T cell count of 240 cells/microL at the time of discontinuation of MAC therapy. Forty-seven subjects remained MAC free, whereas 1 subject developed localized MAC osteomyelitis. The median duration of follow-up while not receiving therapy was 77 weeks, and the incidence of MAC infection was 1.44/100 person-years (95% confidence interval, 0.04-8.01). Withdrawal of anti-MAC therapy appears to be safe in patients who have been treated with a macrolide-based regimen for at least 1 year and have an immunologic response on HAART.