Background. We compared 3 years of antiretroviral therapy with raltegravir or efavirenz as part of a combination regimen in the ongoing STARTMRK study of treatment-naive patients infected with human immunodeficiency virus (HIV). Methods. Eligible patients with HIV-1 RNA (vRNA) levels >5000 copies/mL and without baseline resistance to efavirenz, tenofovir, or emtricitabine were randomized in a double-blind, noninferiority study to receive raltegravir or efavirenz, each combined with tenofovir/emtricitabine. Outcomes included viral suppression, adverse events, and changes from baseline metabolic parameters. Dual energy X-ray absorptiometry scans were obtained on a convenience sample of patients at prespecified time points to assess changes in body fat composition. Results. At week 156 counting noncompleters as failures, 212 (75.4%) of 281 versus 192 (68.1%) of 282 had vRNA levels <50 copies/mL in the raltegravir and efavirenz groups, respectively [Δ (95% CI) = 7.3% (-0.2, 14.7), noninferiority P <. 001]. Mean changes from baseline CD4 count were 332 and 295 cells/mm3 in the raltegravir and efavirenz arms, respectively [Δ (95% CI) = 37 (4, 69)]. Consistent virologic and immunologic efficacy was maintained across prespecified demographic and baseline prognostic subgroups for both treatment groups. Fewer drug-related clinical adverse events (49% vs 80%; P <. 001) occurred in raltegravir than efavirenz recipients, with discontinuations due to adverse events in 5% and 7%, respectively. Elevations in fasting lipid levels (including LDL-and HDL-cholesterol) were consistently lower in the raltegravir than efavirenz group (P <. 005). Fat gain was 19% in 25 raltegravir recipients and 31% in 32 efavirenz recipients at week 156. Conclusions. When combined with tenofovir/emtricitabine in treatment-naive patients, raltegravir produced durable viral suppression and immune restoration that was at least equivalent to efavirenz through 156 weeks of therapy. Both regimens were well tolerated, but raltegravir was associated with fewer drug-related clinical adverse events and smaller elevations in lipid levels. © 2011 The Author.