Background: Adherence is a fundamental determinant of antiretroviral therapy (ART) success. Some programs require patients to identify a support partner to enhance adherence. However, evidence supporting this strategy’s effectiveness is lacking. Methodology: We conducted a 1:1 randomized trial of a treatment partner-based intervention compared with usual care for second-line ART at 9 low and middle income sites. Eligible patients were failing a non-nucleoside-based first-line regimen with HIV RNA >1000 copies/mL and had a willing support partner. The second-line regimen included lopinavir/ritonavir (400/100 mg) twice daily and emtricitabine/tenofovir disoproxil fumarate (200/300 mg) once daily. The intervention included one standardized partner training session focused on adherence support strategies and the partner observing and documenting one ART dose daily > 5 days/week for 24 weeks. Experimental partners were trained to implement adherence strategies and/or telephone the site for help if they observed non-adherence. Control partners received general HIV health information at entry with no additional support. The primary outcome was virologic failure (confirmed HIV RNA >400 copies/mL) at or prior to week 48; secondary outcomes included virologic failure at 24 weeks and adherence measured using electronic monitors (MEMS) summarized as % of doses taken quarterly (Q). The proportion of participants with failure was compared between groups using Fisher’s exact tests. Median adherence per quarter was compared using Wilcoxon rank sum tests. Results: We enrolled 129 patients in the intervention and 128 in usual care, 130 (51%) males, 204 (79%) of African origin, 52 (20%) Latino, with a median age 38 years (interquartile range 33-45). Partners were parents (57, 22%), spouses (55, 21%), siblings (50, 19%), friends (41, 16%), and other (54, 21%). The proportions experiencing virologic failure at week 24 were 19% in the experimental arm vs.13% in the control arm, (p=0.31) and at week 48 were 26% in the experimental arm vs. 18% in the control arm, (p=0.13). Adherence by MEMS was similarly high [Q1: 96% vs. 96%, Q2: 91% vs. 94%, Q3: 90% vs. 93%, Q4: 91% vs. 93%] in experimental and control arms, respectively (all p>0.3). Conclusions: We found no intervention effect on virologic responses or adherence. Potential reasons include control partners providing sufficient support without specific training, our single session being insufficient to significantly enhance skills, the study visits serving as a co-intervention that resulted in high adherence in the controls, or selection bias by enrolling highly adherent motivated volunteers. Continuing work is needed to identify interventions that can improve adherence in persons failing first line therapy.