A prospective, randomized pilot trial was conducted in naive patients comparing three different combinations: zidovudine+lamivudine+lopinavir/ritonavir (arm A) versus tenofovir+lamivudine+efavirenz (arm B) versus tenofovir+didanosine+efavirenz (arm C). HIV-RNA slope (days 1, 3, 7, 14 and 28) was slower in arm C with respect to arm B (P < 0.0001). Seven out of eight patients (87.5%) reached undetectable HIV-RNA by week 28 in arm A, 10/10 (100%) in arm B and 6/10 (60%) in arm C. Among arm C patients who failed at week 4, one HIV isolate showed 67N and 219Q, and another one showed 210F and 215D substitutions in the HIV reverse transcriptase gene at baseline, respectively. Non-nucleoside reverse transcriptase inhibitor resistance-related mutations appeared first, followed by 65R mutations in all cases. Efavirenz AUC(0-24) values were lower in arm C with respect to arm B, especially in patients who failed early. A high virological failure rate after tenofovir+didanosine+efavirenz correlated with a slower HIV-RNA decrease and a peculiar accumulation of resistance mutations. A constellation of factors could be correlated with early failure events in patients receiving this combination such as resistance mutations or polymorphisms present at baseline, low CD4+ T-cell count or advanced disease and unexpectedly low efavirenz plasma levels.