From December 2005 through August 2008, we provided community-based accompaniment with supervised antiretroviral therapy (CASA) to impoverished individuals starting highly active antiretroviral therapy. Adherence support was provided for 18 months by a community-based team comprised of several nurses and two types of community health workers: field supervisors and directly observed therapy (DOT) volunteers. To complement our quantitative data collection in 2008 using purposive sampling, we conducted two gender-mixed focus group discussions with 13 CASA patient participants and 13 DOT volunteers from Lima, Peru to identify the mediating mechanisms by which CASA improved well-being, and to understand the benefits of the intervention, as perceived by these individuals. Using standard qualitative methods for the review and analysis of transcripts and interview notes, we identified central themes and developed a coding scheme for categorising participants’ statements. Two individuals blinded to each other’s coding, coded interview transcripts for theme and content from which a third reviewer compared their coding to arbitrate discrepancies. Additional domains were added if necessary and all domains were integrated into a theoretical scheme. Among the forms of support delivered by the CASA team, DOT volunteers reported emotional support, instrumental support, directly observed therapy, building trust, education, advocacy, exercise of moral authority and preparation for transition off CASA support. CASA participants described outcomes of improved adherence, ability to resume social roles, increased self-efficacy, hopefulness, changes in non-HIV-related behaviour, reduced internalised and externalised stigma, as well as ability to disclose. Both sets of focus group participants highlighted remaining challenges after completion of CASA support: stigma in the community, difficulties achieving economic recovery and persistent barriers to health services. Based on our prior quantitative and qualitative outcomes reported here, we argue that DOT of highly active antiretroviral therapy could be designed to optimise psychosocial recovery during the period of DOT.