Directly observed therapy, which has been successful in the treatment of tuberculosis, is being adapted for the treatment of HIV to decrease long-term morbidity and mortality. We describe the experiences of 69 HIV-infected individuals who were enrolled in a community-based modified directly observed therapy (MDOT) program. Participants were referred by their primary care physicians based on nonadherence to antiretroviral therapy, and/or active substance use. A near-peer outreach worker initially delivered medications to participants 5 to 7 days per week, with visits subsequently tapered to 1 to 3 days per week after 3 or more months. Questionnaires were completed and laboratory values were obtained at baseline, 1 month, and every 3 months after enrollment. At enrollment, 96% of participants had a history of substance use, 71% had a history of incarceration, and 93% were experienced with highly active antiretroviral therapy (HAART). At the time of their 6-month assessment visit, 31 of 69 participants were receiving observed therapy visits. The median baseline plasma viral load (PVL) was 4.8 log, and the median individual change in PVL from baseline to 6 months among participants receiving MDOT was a decrease of 2.7 log. Reasons why participants were not receiving visits included medication holidays, hospitalization or assisted living, incarceration, discontinuation of program involvement, and death. These results support that MDOT should be included in the spectrum of options available to enhance adherence to HAART among patients who are unsuccessful with self-administration of their medications.