Antiretroviral therapy (ART) dramatically reduces mortality due to HIV. Poor adherence to medication is a major barrier to successful treatment. High rates of mental illness, substance use, and unstable housing make adherence particularly problematic in the urban poor. We propose to carry out a three arm, randomized controlled study of two interventions aimed at improving adherence to ART in the urban poor: 1) Adherence Case Management (ACM) and 2) Modified Directly Observed Therapy (MDOT). 336 homeless and marginally housed HIV individuals on either once-daily or twice-daily combination antiretroviral therapy with detectable viral loads and 430% pill count adherence will be randomized to each condition (1 12 per condition). The ACM intervention will be delivered as part of an ongoing adherence support program funded by the San Francisco Department of Public Health. The MDOT program will be operated at our current storefront research site in the Tenderloin District of San Francisco. Adherence to HIV antiretroviral therapy will be measured by unannounced pill count and an audio-CAS1 patient interview method. We will investigate the impact of these two interventions on four specific areas: 1) adherence to antiretroviral therapy, 2) changes in HIV- 1 viral load, 3) development of viral resistance, and 4) cost effectiveness of the interventions. Outcomes will be assessed quarterly during the 6-month intervention and during a 6 month post-intervention period to examine the durability of the intervention effect.