Background: The week 48 primary analysis of the ENCORE1 trial established the virological non-inferiority and safety of efavirenz 400 mg compared with the standard 600 mg dose, combined with tenofovir and emtricitabine, as first-line HIV therapy. This 96-week follow-up of the trial assesses the durability of efficacy and safety of this treatment over 96 weeks. Methods: ENCORE1 was a double-blind, placebo-controlled, non-inferiority trial done at 38 clinical sites in 13 countries. HIV-infected adult patients (≥16 years of age) with no previous antiretroviral therapy, a CD4 cell count of 50-500 cells per μL, and plasma HIV-1 viral load of at least 1000 copies per mL were randomly assigned (1:1) by an electronic case report form to receive fixed-dose daily tenofovir 300 mg and emtricitabine 200 mg plus efavirenz either 400 mg daily or 600 mg daily. Participants, physicians, and all other trial staff were masked to treatment assignment. Randomisation was stratified by HIV-1 viral load at baseline (≤ or >100 000 copies per mL). The primary endpoint was the difference in the proportions of patients in the two treatment groups with a plasma HIV-1 viral load below 200 copies per mL at week 96. Treatment groups were deemed to be non-inferior if the lower limit of the 95% CI for the difference in viral load was above -10% by modified intention-to-treat analysis. Non-inferiority was assessed in the modified intention-to-treat, per-protocol, and non-completer=failure (NC=F) populations. Adverse events and serious adverse events were summarised by treatment group. This study is registered with ClinicalTrials.gov, number NCT01011413. Findings: Between Aug 24, 2011, and March 19, 2012, 636 eligible participants were enrolled and randomly assigned to the two treatment groups (324 to efavirenz 400 mg and 312 to efavirenz 600 mg). The intention-to-treat population who received at least one dose of study drug comprised 630 patients: 321 in the efavirenz 400 mg group and 309 in the efavirenz 600 mg group. 585 patients (93%; 299 in the efavirenz 400 mg group and 286 in the 600 mg group) completed 96 weeks of follow-up. At 96 weeks, 289 (90·0%) of 321 patients in the efavirenz 400 mg group and 280 (90·6%) of 309 in the efavirenz 600 mg group had a plasma HIV-1 viral load less than 200 copies per mL (difference -0·6, 95% CI -5·2 to 4·0; p=0·72), which suggests continued non-inferiority of the lower efavirenz dose. Non-inferiority was recorded for thresholds of less than 50 and less than 400 copies per mL, irrespective of baseline plasma viral load. Adverse events were reported by 291 (91%) of 321 patients in the efavirenz 400 mg group and by 285 (92%) of 309 in the 600 mg group (p=0·48). The proportions of patients reporting an adverse event that was definitely or probably related to efavirenz were 126 (39%) for efavirenz 400 mg and 148 (48%) for efavirenz 600 mg (p=0·03). The number of patients who reported serious adverse events did not differ between the groups (p=0·20). Interpretation: Our findings confirm that efavirenz 400 mg is non-inferior to the standard dose of 600 mg in combination with tenofovir and emtricitabine as initial HIV therapy over 96 weeks. Fewer efavirenz-related adverse events were reported with the 400 mg efavirenz dose than with the 600 mg dose. These findings support the routine use of efavirenz 400 mg. The coadministration of rifampicin and efavirenz 400 mg needs further investigation. Funding: Bill & Melinda Gates Foundation, and UNSW Australia.