OBJECTIVES: Recommended treatment regimens for HIV-1-infected patients combine antivirals from different therapeutic classes, usually a two-drug backbone (TDF/FTC or ABC/3TC) with a third agent from another therapeutic class. We sought to understand the clinical evidence differentiating these backbones in treatment-naïve patients. METHODS: A SLR identified randomised controlled trials (RCTs) of TDF/FTC or ABC/3TC plus third agents grouped by class (protease inhibitors [PI], non-nucleoside reverse transcriptase inhibitors [NNRTI] and integrase strand transfer inhibitors [INSTI]) in HIV-1 infection. MEDLINE, EMBASE and the Cochrane Library were searched in March 2014. Bayesian NMAs of RCTs in were run for virologic response (VR) and all-cause discontinuation at Week 48 (Wk48) and 96 (Wk96). Inconsistency and the effect of baseline characteristics were also assessed using unrelated mean-effects models and meta-regression, respectively. RESULTS: Of 1,093 citations retrieved, 243 citations were included in the SLR, reporting 18 RCTs that informed at least one network. In the NMA, fixed-effect models represented a better fit for VR data, whereas random-effects models fitted the all-cause discontinuation data best. With NNRTIs, TDF/FTC was associated with significantly higher odds of VR than ABC/3TC at Wk48 (OR 1.32 [95%CrI 1.05, 1.65]) and Wk96 (OR 1.29 [95%CrI 1.03, 1.61]). With INSTIs, TDF/FTC had a significantly higher odds of VR at Wk96 compared with ABC/3TC (OR 1.46 [95%CrI 1.04, 2.04]). No statistically significant differences in VR were found between the backbones with PIs. No statistically significant differences in all-cause discontinuation at Wk48 or Wk96 were observed between the backbones with any class of third agent. Networks showed little inconsistency, and baseline characteristics did not have any significant effect on results. CONCLUSIONS: TDF/FTC was associated with statistically significant VR benefits compared with ABC/3TC with NNRTIs at both Wk48 and Wk96 and with INSTIs at Wk96, and no statistically significant effect was seen with respect to all-cause discontinuation.