BackgroundGuidelines recommend both protease inhibitor (PI)- and non-nucleoside reverse transcriptase inhibitor (NNRTI)-containing regimens for initial therapy in HIV-positive individuals whilst clinical trial data comparing treatment options remain limited.ObjectiveTo assess whether drug selection (PI versus NNRTI) in antiretroviral-naive patients is related to virological response at 6 months within a clinical cohort.DesignDatabases from two large clinics were used to identify all treatment-naive patients initiating highly active antiretroviral therapy (PI/ two PI or NNRTI). Statistical models determined the likelihood of suppressing HIV viral load < 500 copies/ml, the risk of treatment failure by 6 months, and factors associated with treatment success.ResultsOf 1109 potentially eligible patients 888 met study criteria and were included; 484 were prescribed a PI (40% indinavir, 41% nelfinavir) and 404 were prescribed NNRTI (40% efavirenz, 60% nevirapine). Three treatment arms were compared: efavirenz versus nevirapine versus PI. After stratification by year and centre and adjustment for baseline variables, only treatment group and baseline viral load remained significantly associated with virological suppression at 6 months. Patients on efavirenz were significantly more likely to achieve an undetectable viral load than those on PI or nevirapine. The relative hazard for nevirapine was 0.77 (95% confidence interval, 0.61–0.96,P= 0.02) and that for PI was 0.74 (95% confidence interval, 0.58–0.94,P= 0.01). Efavirenz also performed better in the analysis of treatment failure at 6 months.ConclusionAlthough observational cohort data may be susceptible to significant bias, this study suggests a better initial virological response for efavirenz compared to either nevirapine or the PI. Clinical trial data is required to confirm these findings.